determination of natural radioactivity concentrations in soil
TRANSCRIPT
Determination of Natural Radioactivity Concentrations in Soil a comparative study of
Windows and Full Spectrum Analysis
by
Katse Piet Maphoto
Thesis submitted in partial fulfillment of the
requirements for the MSc degree in Physics at the
University of the Western Cape
August 2004
Supervisor Prof R Lindsay
Co-supervisor Dr RT Newman (iThemba LABS)
Declaration
I the undersigned declare that the work contained in this thesis is my own original work and
has not previously in its entirety or in part been submitted at any university for a degree
Signature
Date
ii
Acknowledgements
I am greatly indebted to the following people who made this thesis possible
Prof Robbie Lindsay supervisor for the help in financing my studies and guiding me every
step of the way especially with challenging aspects such as programming
Dr Richard Thomas Newman co-supervisor for the guidance and suggestions especially
during the writing up of the thesis
iThemba LABS for financing and granting me the opportunity to do my study with them
and to all the staff from the library to the Human Resource for their support
To my parents Magate le Baboloki for always being on my side whenever I needed them
and for their love and patience and my Family for their support and endless love
Environmental Radioactivity Group Dawid de Villiers Angelo Joseph Raphael Damon
and Lerato Sedumedi (former Technical Officer) for their helpful discussions
To Prof Rob de Meijer from the Kernfysisch Versneller Instituut (KVI) in the Netherlands
for helpful discussions and suggestions
To Peane Maleka former masters student at the iThemba LABS for introducing me to the
basic concepts of experimental methodologies
UWC Physics department for making me feel at home when I felt like a stranger and all my
colleagues for their encouragement
Finally thanks to the Almighty God the Creator of the Universe the Alpha and Omega for
keeping me alive and kicking all the way
iii
Determination of Natural Radioactivity Concentrations in Soil a comparative study of
Windows and Full Spectrum Analysis
Katse Piet Maphoto
Abstract
In this study two methods of analysing activity concentrations of natural radionuclides (238U 232Th and 40K) in soil are critically compared These are the Window Analysis (WA) and Full
Spectrum Analysis (FSA) In the usual WA method the activity concentrations are determined
from the net counts of the windows set around individual γ-ray peaks associated with the
decay of 238U 232Th and 40K In the FSA method the full energy spectrum is considered and
the measured spectrum is described as the sum of the three standard spectra (associated with 238U 232Th and 40K respectively) each multiplied by an unknown concentration The
concentrations are determined from the FSA and correspond to the activity concentrations of 238U 232Th and 40K in the soil The standard spectra derived from separate calibration
measurements using the HPGe detector represents the response of the HPGe to a Marinelli
sample beaker containing an activity concentration of 1 Bqkg
A χ2 -minimisation procedure (for the FSA method) is used to extract the accurate activity
concentrations The results of this technique (FSA) were obtained and compared with the
conventional (WA) method of analysis Although a good correlation was found between
results from these methods the disadvantage was with the self-absorption effects which result
into poor fit (especially at the continuum) This proved to have had an impact on the measured
results for FSA This phenomenon was carefully studied and a recommendation is made in the
conclusion
Five samples having a range of activity concentrations (relative and absolute) and densities
were analyzed as part of this study In one case (that for the (stearic acid + 232Th) sample) the
absolute activity concentration of 232Th in the sample was known The samples were measured
iv
for longer times (about 8 to 10 hours) and short times (about 1 hour) Activity concentrations
were then calculated from each spectrum using WA and FSA For the long measurements the
ratio of WA-determined concentrations to FSA-determined concentrations varied between
about 10 and 13 for 238U between 10 and 16 for 232Th and between 097 and 10 for 40K
The corresponding ratio for short measurements varied between 10 and 12 for 238U between
07 and 11 for 232Th and between 09 and 10 for 40K
This is with the exception of the low activity sample of the West coast sample which gave the
ratios ranging from 097 to 16 in the case of long measurement and 14 to 74 in case of the
short measurements
In the case of the (stearic acid + 232Th) sample the WA and FSA determined activity
concentrations deviated from the expected concentrations by 9 and 21 respectively
This study demonstrated that the FSA technique could be useful for extracting activity
concentrations from high-resolution gamma ray spectra in a relatively straightforward and
convenient way provided that proper consideration is given to effects associated with sample
volume and density in relation to the sources used to generate standard spectra
v
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
Declaration
I the undersigned declare that the work contained in this thesis is my own original work and
has not previously in its entirety or in part been submitted at any university for a degree
Signature
Date
ii
Acknowledgements
I am greatly indebted to the following people who made this thesis possible
Prof Robbie Lindsay supervisor for the help in financing my studies and guiding me every
step of the way especially with challenging aspects such as programming
Dr Richard Thomas Newman co-supervisor for the guidance and suggestions especially
during the writing up of the thesis
iThemba LABS for financing and granting me the opportunity to do my study with them
and to all the staff from the library to the Human Resource for their support
To my parents Magate le Baboloki for always being on my side whenever I needed them
and for their love and patience and my Family for their support and endless love
Environmental Radioactivity Group Dawid de Villiers Angelo Joseph Raphael Damon
and Lerato Sedumedi (former Technical Officer) for their helpful discussions
To Prof Rob de Meijer from the Kernfysisch Versneller Instituut (KVI) in the Netherlands
for helpful discussions and suggestions
To Peane Maleka former masters student at the iThemba LABS for introducing me to the
basic concepts of experimental methodologies
UWC Physics department for making me feel at home when I felt like a stranger and all my
colleagues for their encouragement
Finally thanks to the Almighty God the Creator of the Universe the Alpha and Omega for
keeping me alive and kicking all the way
iii
Determination of Natural Radioactivity Concentrations in Soil a comparative study of
Windows and Full Spectrum Analysis
Katse Piet Maphoto
Abstract
In this study two methods of analysing activity concentrations of natural radionuclides (238U 232Th and 40K) in soil are critically compared These are the Window Analysis (WA) and Full
Spectrum Analysis (FSA) In the usual WA method the activity concentrations are determined
from the net counts of the windows set around individual γ-ray peaks associated with the
decay of 238U 232Th and 40K In the FSA method the full energy spectrum is considered and
the measured spectrum is described as the sum of the three standard spectra (associated with 238U 232Th and 40K respectively) each multiplied by an unknown concentration The
concentrations are determined from the FSA and correspond to the activity concentrations of 238U 232Th and 40K in the soil The standard spectra derived from separate calibration
measurements using the HPGe detector represents the response of the HPGe to a Marinelli
sample beaker containing an activity concentration of 1 Bqkg
A χ2 -minimisation procedure (for the FSA method) is used to extract the accurate activity
concentrations The results of this technique (FSA) were obtained and compared with the
conventional (WA) method of analysis Although a good correlation was found between
results from these methods the disadvantage was with the self-absorption effects which result
into poor fit (especially at the continuum) This proved to have had an impact on the measured
results for FSA This phenomenon was carefully studied and a recommendation is made in the
conclusion
Five samples having a range of activity concentrations (relative and absolute) and densities
were analyzed as part of this study In one case (that for the (stearic acid + 232Th) sample) the
absolute activity concentration of 232Th in the sample was known The samples were measured
iv
for longer times (about 8 to 10 hours) and short times (about 1 hour) Activity concentrations
were then calculated from each spectrum using WA and FSA For the long measurements the
ratio of WA-determined concentrations to FSA-determined concentrations varied between
about 10 and 13 for 238U between 10 and 16 for 232Th and between 097 and 10 for 40K
The corresponding ratio for short measurements varied between 10 and 12 for 238U between
07 and 11 for 232Th and between 09 and 10 for 40K
This is with the exception of the low activity sample of the West coast sample which gave the
ratios ranging from 097 to 16 in the case of long measurement and 14 to 74 in case of the
short measurements
In the case of the (stearic acid + 232Th) sample the WA and FSA determined activity
concentrations deviated from the expected concentrations by 9 and 21 respectively
This study demonstrated that the FSA technique could be useful for extracting activity
concentrations from high-resolution gamma ray spectra in a relatively straightforward and
convenient way provided that proper consideration is given to effects associated with sample
volume and density in relation to the sources used to generate standard spectra
v
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
Acknowledgements
I am greatly indebted to the following people who made this thesis possible
Prof Robbie Lindsay supervisor for the help in financing my studies and guiding me every
step of the way especially with challenging aspects such as programming
Dr Richard Thomas Newman co-supervisor for the guidance and suggestions especially
during the writing up of the thesis
iThemba LABS for financing and granting me the opportunity to do my study with them
and to all the staff from the library to the Human Resource for their support
To my parents Magate le Baboloki for always being on my side whenever I needed them
and for their love and patience and my Family for their support and endless love
Environmental Radioactivity Group Dawid de Villiers Angelo Joseph Raphael Damon
and Lerato Sedumedi (former Technical Officer) for their helpful discussions
To Prof Rob de Meijer from the Kernfysisch Versneller Instituut (KVI) in the Netherlands
for helpful discussions and suggestions
To Peane Maleka former masters student at the iThemba LABS for introducing me to the
basic concepts of experimental methodologies
UWC Physics department for making me feel at home when I felt like a stranger and all my
colleagues for their encouragement
Finally thanks to the Almighty God the Creator of the Universe the Alpha and Omega for
keeping me alive and kicking all the way
iii
Determination of Natural Radioactivity Concentrations in Soil a comparative study of
Windows and Full Spectrum Analysis
Katse Piet Maphoto
Abstract
In this study two methods of analysing activity concentrations of natural radionuclides (238U 232Th and 40K) in soil are critically compared These are the Window Analysis (WA) and Full
Spectrum Analysis (FSA) In the usual WA method the activity concentrations are determined
from the net counts of the windows set around individual γ-ray peaks associated with the
decay of 238U 232Th and 40K In the FSA method the full energy spectrum is considered and
the measured spectrum is described as the sum of the three standard spectra (associated with 238U 232Th and 40K respectively) each multiplied by an unknown concentration The
concentrations are determined from the FSA and correspond to the activity concentrations of 238U 232Th and 40K in the soil The standard spectra derived from separate calibration
measurements using the HPGe detector represents the response of the HPGe to a Marinelli
sample beaker containing an activity concentration of 1 Bqkg
A χ2 -minimisation procedure (for the FSA method) is used to extract the accurate activity
concentrations The results of this technique (FSA) were obtained and compared with the
conventional (WA) method of analysis Although a good correlation was found between
results from these methods the disadvantage was with the self-absorption effects which result
into poor fit (especially at the continuum) This proved to have had an impact on the measured
results for FSA This phenomenon was carefully studied and a recommendation is made in the
conclusion
Five samples having a range of activity concentrations (relative and absolute) and densities
were analyzed as part of this study In one case (that for the (stearic acid + 232Th) sample) the
absolute activity concentration of 232Th in the sample was known The samples were measured
iv
for longer times (about 8 to 10 hours) and short times (about 1 hour) Activity concentrations
were then calculated from each spectrum using WA and FSA For the long measurements the
ratio of WA-determined concentrations to FSA-determined concentrations varied between
about 10 and 13 for 238U between 10 and 16 for 232Th and between 097 and 10 for 40K
The corresponding ratio for short measurements varied between 10 and 12 for 238U between
07 and 11 for 232Th and between 09 and 10 for 40K
This is with the exception of the low activity sample of the West coast sample which gave the
ratios ranging from 097 to 16 in the case of long measurement and 14 to 74 in case of the
short measurements
In the case of the (stearic acid + 232Th) sample the WA and FSA determined activity
concentrations deviated from the expected concentrations by 9 and 21 respectively
This study demonstrated that the FSA technique could be useful for extracting activity
concentrations from high-resolution gamma ray spectra in a relatively straightforward and
convenient way provided that proper consideration is given to effects associated with sample
volume and density in relation to the sources used to generate standard spectra
v
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
Determination of Natural Radioactivity Concentrations in Soil a comparative study of
Windows and Full Spectrum Analysis
Katse Piet Maphoto
Abstract
In this study two methods of analysing activity concentrations of natural radionuclides (238U 232Th and 40K) in soil are critically compared These are the Window Analysis (WA) and Full
Spectrum Analysis (FSA) In the usual WA method the activity concentrations are determined
from the net counts of the windows set around individual γ-ray peaks associated with the
decay of 238U 232Th and 40K In the FSA method the full energy spectrum is considered and
the measured spectrum is described as the sum of the three standard spectra (associated with 238U 232Th and 40K respectively) each multiplied by an unknown concentration The
concentrations are determined from the FSA and correspond to the activity concentrations of 238U 232Th and 40K in the soil The standard spectra derived from separate calibration
measurements using the HPGe detector represents the response of the HPGe to a Marinelli
sample beaker containing an activity concentration of 1 Bqkg
A χ2 -minimisation procedure (for the FSA method) is used to extract the accurate activity
concentrations The results of this technique (FSA) were obtained and compared with the
conventional (WA) method of analysis Although a good correlation was found between
results from these methods the disadvantage was with the self-absorption effects which result
into poor fit (especially at the continuum) This proved to have had an impact on the measured
results for FSA This phenomenon was carefully studied and a recommendation is made in the
conclusion
Five samples having a range of activity concentrations (relative and absolute) and densities
were analyzed as part of this study In one case (that for the (stearic acid + 232Th) sample) the
absolute activity concentration of 232Th in the sample was known The samples were measured
iv
for longer times (about 8 to 10 hours) and short times (about 1 hour) Activity concentrations
were then calculated from each spectrum using WA and FSA For the long measurements the
ratio of WA-determined concentrations to FSA-determined concentrations varied between
about 10 and 13 for 238U between 10 and 16 for 232Th and between 097 and 10 for 40K
The corresponding ratio for short measurements varied between 10 and 12 for 238U between
07 and 11 for 232Th and between 09 and 10 for 40K
This is with the exception of the low activity sample of the West coast sample which gave the
ratios ranging from 097 to 16 in the case of long measurement and 14 to 74 in case of the
short measurements
In the case of the (stearic acid + 232Th) sample the WA and FSA determined activity
concentrations deviated from the expected concentrations by 9 and 21 respectively
This study demonstrated that the FSA technique could be useful for extracting activity
concentrations from high-resolution gamma ray spectra in a relatively straightforward and
convenient way provided that proper consideration is given to effects associated with sample
volume and density in relation to the sources used to generate standard spectra
v
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
for longer times (about 8 to 10 hours) and short times (about 1 hour) Activity concentrations
were then calculated from each spectrum using WA and FSA For the long measurements the
ratio of WA-determined concentrations to FSA-determined concentrations varied between
about 10 and 13 for 238U between 10 and 16 for 232Th and between 097 and 10 for 40K
The corresponding ratio for short measurements varied between 10 and 12 for 238U between
07 and 11 for 232Th and between 09 and 10 for 40K
This is with the exception of the low activity sample of the West coast sample which gave the
ratios ranging from 097 to 16 in the case of long measurement and 14 to 74 in case of the
short measurements
In the case of the (stearic acid + 232Th) sample the WA and FSA determined activity
concentrations deviated from the expected concentrations by 9 and 21 respectively
This study demonstrated that the FSA technique could be useful for extracting activity
concentrations from high-resolution gamma ray spectra in a relatively straightforward and
convenient way provided that proper consideration is given to effects associated with sample
volume and density in relation to the sources used to generate standard spectra
v
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
Table of Contents
CHAPTER 1 Introduction 1
11 The atomic nuclei and radioactivity overview2
111 Radioactive decay3
112 Decay modes3
113 Decay rates5
12 Types of environmental radioactivity6
13 Review of primordial radioactivity7
131 Decay series of natural radionuclide7
14 Literature review natural radionuclide activity concentrations in soil12
141 Methods used to determine activity concentrations in soil13
142 Interactions of gamma rays with matter14
1421 Photoelectric absorption14
1422 Compton scattering15
1423 Pair production17
1424 Attenuation Coefficients18
143 Examples of gamma ray spectrometry systems19
vi
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
15 The motivation for this study21
16 The aim and scope for this study23
17 Thesis outline24
CHAPTER 2 Experimental Methods25
21 The HPGe gamma-ray detection facility25
211 Physical layout26
212 HPGe technical specifications28
213 Electronics28
214 Figure of merit34
22 Sample Selection Preparation and Measurement41
23 Reference Sources43
24 Data Acquisition44
CHAPTER 3 Data Analysis46
31 Window Analysis46
311 Determination of γ-ray detection efficiency49
312 Treatment of uncertainties in WA 62
32 Full Spectrum Analysis64
vii
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
321 Derived standard spectra64
322 Chi-square minimisation technique70
323 Treatment of uncertainties for FSA86
CHAPTER 4 Results and Discussion88
41 WA Results88
42 FSA Results97
43 Comparison of WA and FSA Results99
44 Discussion of WA Results110
45 Discussion of FSA Results111
CHAPTER 5 Conclusion119
51 Outlook119
APPENDICES
A-1 Some Formulae for Combining uncertainties121
A-2 Means and Standard deviation122
B FORTRAN program125
C Background Spectrum127
D Self-absorption effects128
viii
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
E Ratios of activity concentrations133
ix
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
List of Tables
1-1 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and
1 m deep 12
1-2 Characteristic radiometric fingerprints of sand and mud 13
1-3 Radiometric fingerprint given as activity concentrations (Bqkg-1) associations with
heavy minerals in South Africa 22
2-1 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines
associated with the decay of 40K and the series of 232Th and 238U used for energy
calibrations 35
2-2 A table of counts in the chosen region of interest with number of channels along the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak 40
2-3 The table of samples collected at different sites 41
2-4 The table shows data on the reference material used 44
2-5 Spectral information on reference sources 45
2-6 Spectral information on Samples and background 45
3-1 The gamma ray lines used and their associated branching ratios 52
4-1 The activity concentration results for the Kloof sample number 6 by WA method 88
x
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
4-2 The activity concentration results (long measurement) for the Westonaria sample
number 17A by WA method 89
4-3 The activity concentration results (long measurement) for the West Coast sample
number 3 by WA method 90
4-4 The activity concentration results for (long measurement) the Brick clay sample
number 6 by WA method 91
4-5 The activity concentration results for (long measurement) the thorium + stearic acid
sample number 5 by WA method 92
4-6 The activity concentration results (short measurement) for the Kloof sample number 6
by WA method 93
4-7 The activity concentration results (short measurement) for the Westonaria sample
number 17A by WA method 94
4-8 The activity concentration results (short measurement) for the West Coast sample
number 3 by WA method 95
4-9 The activity concentration results (short measurement) for the Brick clay sample
number 6 by WA method 96
4-10 The FSA results (8-10 hours measurements) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 97
xi
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released
4-11 The FSA result (one hour measurement) uncertainties and the associated chi-square
per degree of freedom obtained using a cut off energy of 300 keV for the samples
indicated 98
4-12 The activity concentration from the two methods of analysis for long measurements 99
4-13 The activity concentration from the two methods of analysis for short time
measurement 100
4-14 Advantages and disadvantages for the two methods including the optimum
measurement times required for the activity concentration determination 108
4-15 The results for sample 5 a high thorium content sample 109
A-1 The table shows some of the equations used in the calculation of the uncertainty ∆R
121
D-1 Results of calculations of the effective thickness t for two Marinelli beaker volumes 130
D-2 A table of the mass attenuation coefficients and the transmission factors 131
D-3 The dimensions of the full and half-full Marinelli beaker 132
E-1 The ratios of the activity concentrations (WAFSA) for samples used in the study
the ratios are shown for both short and long measurement 133
xii
LIST OF FIGURES
1-1 40K decays by β- and electron capture to 40Ar and 40Ca 8
1-2 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective
nuclides are indicated 9
1-3 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective
nuclides are indicated 10
1-4 The schematic representation of the mechanism of photoelectric absorption where a γ-
ray with energy Eγ interacts with a bound electron 14
1-5 The diagrammatic representation of emission of fluorescent X-rays 15
1-6 The diagrammatic illustration of mechanism of Compton scattering 16
1-7 The diagrammatic illustration of pair production 17
1-8 The linear attenuation coefficient of germanium and its component parts on a log-log
scale 19
1-9 Application of Radiometric methods in different fields 21
2-1 The picture showing the setup of the Environmental Radioactivity Laboratory at
iThemba LABS 26
2-2 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector 27
xiii
2-3 The open top view showing the detector (A) and a Marinelli beaker on top of the
detector (B) 28
24 Schematic diagram illustrating the electronic setup used to acquire the HPGe data 29
2-5 Coaxial Ge Detector Cross Section 30
2-6 A diagram showing a typical semiconducter with band gap Eg 30
2-7 A typical germanium detector cryostat and liquid nitrogen reservoir (dewar) 31
2-8 Basic architecture of an MCA 33
2-9 The energy calibration spectrum showing the lines (labelled) used to perform energy
calibrations a mixture of 40K 232Th and 238U was used as a source 36
2-10 Energy calibration of the spiked material points and the line of best fit 37
2-11 A Full Width at Half Maximum calibration graph with a line of best fit 39
2-12 A spectrum of 60Co for peak-to-Compton ratio determination 40
2-13 Photo of Marinelli beaker and the design of its geometry the bottom part shows its re-
entrant hole 42
3-1 A graphical representation of the region of interest window set around the 40K peak 48
3-2 A typical representation of a sample spectrum number 6 (Kloof sample) with
superimposed background spectrum on a log scale
3-3 The flow chart showing the steps followed in constructing the absolute efficien
curve
xiv
49
cy
51
3-4 The spectrum of Kloof sand sample showing the lines used during detection efficiency
calibration 53
3-5 Fit of relative efficiency (with parameters a amp b generated) as determined from lines
associated with the decay of 238U (for a Kloof sample number 6) 54
3-6 Fit of relative efficiency with parameters a amp b generated (for Kloof sample) as
determined from lines associated with the decay of 232Th 57
3-7 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Kloof sample) 58
3-8 A relative efficiency curve for the sample number 6 (Kloof sample) 58
3-9 A typical absolute efficiency versus energy graph for various selected lines associated
with nuclides 238U 40K and 232Th from sample number 6 (Kloof sample) 59
3-10 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Westonaria sand sample) 60
311 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (West coast sand sample) 60
3-12 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (Brick clay) 61
3-13 A fit of relative efficiency data with parameters a amp b generated as determined from
lines associated with 238U and 232Th decay (thorium + stearic sample) 61
3-14 The contents of channels of the measured spectrum S redistributed to the re-binned
spectrum S 65
xv
3-15 Flow chart showing how a standard spectrum was obtained 66
3-16 The background subtracted re-binned standard spectra for uranium 67
3-17 The background subtracted re-binned standard spectra for thorium 68
3-18 The background subtracted re-binned standard spectra for potassium 69
3-19 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria
sample as a function of lower cut-off energy 73
3-20 The background subtracted Kloof (a )and West Coast(b) sample spectra and their fit for
a long measurement (below the 300 keV energy) 74
3-21 The background subtracted Brick clay (c) and thorium sample(d) spectra and their fit for
a long measurement (below the 300 keV energy) 75
322 The background subtracted Kloof sample spectrum for a long measurement and
the fit (for energies between 300 and 1000 keV) 76
3-23 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 1000 and 2000 keV) 77
3-24 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 78
3-25 The background subtracted Kloof sample spectrum for a long measurement and the
fit (for energies between 2000 and 2650 keV) 79
xvi
3-26 The background subtracted Kloof sample spectrum for a short measurement and the
fit (for energies between 500 and 1500 keV) 80
3-27 The background subtracted Kloof sample spectrum for a short measurement and
the fit (for energies between 1500 and 2650 keV) 81
3-28 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 300 and 1000 keV) 82
3-29 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 1000 and 2000 keV) 83
3-30 The background subtracted thorium + stearic acid sample spectrum and the fit for a
long measurement (for energies between 2000 and 2600 keV) 84
3-31 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit 85
3-32 A typical 583 keV peak due to 208Tl ( for thorium + stearic acid sample number 5) with
a fit 85
4-1 The comparison of the three nuclides for Westonaria sand (long measurement) in both
window and FSA analyses 101
4-2 The comparison of the three nuclides for Westonaria sand (short measurement) in
both window and FSA analyses 101
4-3 The percentage uncertainties for activity concentrations as a function of nuclide for
Westonaria sand sample 101
xvii
4-4 The comparison of the three nuclides for Kloof sand (long measurement) in both
window and FSA analyses 102
4-5 The comparison of the three nuclides for Kloof sand (short measurement) in both
window and FSA analyses 102
4-6 The percentage uncertainties for activity concentrations as a function of nuclide for
Kloof sand sample 102
4-7 The comparison of the three nuclides for West Coast sand (long measurement) in both
window and FSA analyses 103
4-8 The comparison of the three nuclides for West Coast sand (short measurement) in both
window and FSA analyses 103
4-9 The percentage uncertainties for activity concentrations as a function of nuclide for West
Coast sand sample 103
4-10 The comparison of the three nuclides for Brick clay (long measurement) in both
window and FSA analyses 104
4-11 The comparison of the three nuclides for Brick clay sand (short measurement) in both
window and FSA analyses 104
4-12 The percentage uncertainties for activity concentrations as a function of nuclide for
Brick clay sample 104
4-13 A graph showing correlation of activity concentration determined using the WAM vs
FSA (on average the two methods agree well within the correlation coefficient of
0932) 105
xviii
4-14 The activity concentration ratios (WAFSA) of uranium thorium and potassium long
measurement for various samples 106
4-15 The activity concentration ratios (WAFSA) of uranium thorium and potassium short
measurement for various samples 107
4-16 Activity concentration of nuclides for the high thorium content sample 109
4-17 Calculated transmission factors at different matrices for a 1000 cm3 Marinelli as a
function of γ-ray energy 113
4-18 The transmission factor for a 500 cm3 Marinelli at different densities with an increase in
energy 114
4-19 The volume effect on the transmission factor for different fillings (with sand of density
16 gcm3 ) of Marinelli as a function of energy 115
4-20 The differences in the transmission factors of (Westonaria sand) with regard to full and
half full Marinelli beaker as a function of energy 117
4-21 A filled Marinelli beaker showing an incoming photon and possibilities of interaction
in both the beaker and detector 118
C-1 The background spectrum of Marinelli beaker full of tap water 127
D-1 The Marinelli beaker with a spherical model at the center 128
xix
C h a p t e r 1
1 Introduction
In 1895 the German physicist Wilhelm Roentgen identified penetrating radiation which
produced fluorescence1 and which he named X-rays In 1896 two months later Henri
Becquerel discovered that penetrating radiation later classified as α β and γ rays were
given off in the radioactive decay of uranium and thus opened a new field of study of
radioactive substances and radiations they emit [Sha72]
Rutherford and Soddy were the first to suggest that radioactive atoms disintegrate into lighter
structures as they emit radiation This suggestion received powerful support with the discovery
that the α-particle was just an ionised atom of the element helium Many of the newly
discovered elements were found in various fractions of uranium ores and this the heaviest
naturally occurring element was soon suspected to be the parent substance [Ral72] It is
presently known that uranium consists naturally of a mixture of 238U (9927) 235U (072)
and 234U (0006) thorium and potassium were also identified as the radioactive parents with
some decay products Refer to Figures 11 12 and 13 for the decay processes of natural
radionuclides 238U 232Th and 40K into their daughter nuclei
Soils are naturally radioactive primarily because of their mineral content The main
radionuclides are 238U 232Th and their decay products and 40K The radioactivity varies from
one soil type to the other depending on the mineral makeup and composition [Dra03] The
objective in the measurement of the radioactivity in soil is to asses the radionuclide
concentrations and these concentrations may be used to characterise substances and relate the
1The emission of electromagnetic radiation especially of visible light stimulated in a substance by the absorption of incident radiation and persisting only as long as the stimulating radiation is continued
1
radiometric properties to physical properties of the material This may be of use in mineral
separation for example
11 The atomic nuclei and radioactivity an overview
This section describes the nature of atoms their building blocks and the nature of the forces
playing a role in it In the fourth century BC the Greek philosopher Democritus believed that
each kind of material could be subdivided into smaller and smaller bits until the limit beyond
which no further division was possible These building blocks of matter invisible to the naked
eye were referred to as atoms This speculation was confirmed by experimental scientists
(Dalton Avagadro Faraday) over 2000 years later Once the classification and kind of atoms
have been studied according to the rules governing their combination in matter by the
chemists the next step was to study the fundamental properties of an atom of various
elements This kind of atomic physics led to the discovery of radioactivity of certain species of
atoms [Kra88] Rutherford then used these radiations of atoms as probes of the atoms
themselves In 1911 he then proposed the existence of the atomic nucleus which was
experimentally confirmed by Geiger and Marsden A new branch of science which is now
called nuclear physics was then born This branch of physics is dedicated to studying matter at
its fundamental level
A nucleus X is made up of Z protons and N neutrons (nucleons) with the notation
with atomic mass number A = Z + N where X is a nuclide symbol The mass of a nucleus is
less than the sum of the individual masses of the protons and neutrons which constitute it
The difference is a measure of the nuclear binding energy which holds the nucleus together
This is the energy required to break it into separate neutrons and protons If the nuclear radius
is R then the corresponding volume (43)πR
NAZ X
3 is found to be proportional to A This
relationship is expressed in inverse form as
R = R0A13 (11)
2
with the value of R0 asymp 12 x 10-15m
The description of the nucleus can be understood with the aid of different models Two of
these are the liquid drop model and the shell model [Bei87 Eva55]
111 Radioactive Decay
Protons are positively charged and repel one another electrically Therefore an excess of
neutrons which produce only an attractive force is required for stability thus leading to N gt Z
for stable nuclei There is a limit to the ability of neutrons to prevent the disruption of the
nucleus by the Coulomb repulsion This phenomenon therefore leads to instability of nuclei
The instability can also be attributed to the unstable configurations due to the filling of
quantum states by the nucleon [Hew85]
112 Decay modes
Because of their instability nuclei decay by α β or γ emissions Many naturally occurring
heavy nuclei with 82 lt Z le 92 decay by α emission in which the parent nucleus loses both
mass and charge (A Z) [Ral72] The α (4He-nucleus) type reaction can be represented as
(12) γα ++rarr minusminus
42
42YX A
ZAZ
where X represents a chemical symbol of the parent atom and Y that of the daughter In some
emissions there is no γ emission
In β- particle emission a neutron (in the nucleus) is converted into a proton electron and an
anti-neutrino
epn νβ ++rarr minus01
11
10 (13)
3
Although free electrons do not exist inside the nucleus a β particle originates there and must
pass the nuclear potential barrier to escape [Ral92] This leads to the equation
eA
ZAZ YX νγβ +++rarr minus+
011 (14)
As in the α particle case the atomic mass number and atomic number are conserved The γ
term allows for the possibility that a daughter nucleus will be formed in an excited state while
some transitions go directly to the ground state The β- particle emission is an example of the
radioactive decay of a nuclide with neutron excess In this case a neutron is converted to a
proton according to equation (13)
The neutron-deficient nuclei emit a positron as they go to stability by
(15) enp νβ ++rarr +01
10
11
and the transformation is now
(16) eA
ZAZ YX νγβ +++rarr +
minus011
The energy of α and γ- radiation is discrete and well defined whilst β emission gives βs with a
continuous spectrum due to the varying amount of energy taken away by the neutrino
In the case of electron capture (EC) the neutron-deficient nuclei must decay by a p rarr n
conversion but have daughter products whose mass is greater than the maximum acceptable
for positron emission Therefore these nuclei can only decay by the capture of one of the
orbital electrons The atomic number is then reduced by one unit due to the negative charge
acquired by the nucleus and thus yielding the same daughter that would have been produced
by the positron emission Figure 11 in section 131 presents the decay scheme of 40K in which
4
the electron capture (EC) is in competition with positron emission in addition to β- decay to 40Ca the decay to 40Ar has two competing modes of decay that is β+ and EC The electron
capture is represented by the type of reaction
(17) eA
ZAZ YeX νγ ++rarr+ minus
minusminus 1
01
113 Decay rates
The radioactive decay rate of a nucleus is measured in terms of the decay constant λ which is
the probability per unit time that a given nucleus will decay and this is related to its
half-life2 Each radioactive nucleus has its own characteristic half-life If there are N radioactive
nuclei in a sample the rate of decay will then be given by
NdtdN λminus= (18)
where the minus sign indicates that N is decreasing with time The solution of this equation is
(19) teNtN λminus= )0()(
with N(0) being the number of nuclei present at t = 0 The half-life t12 may be expressed in
terms of λ from equation (113) as
λ
2ln21 =t (110)
The activity A is defined as the number of decaying nuclei per unit time and it can be
represented by
2 The time needed for half of the radioactive nuclei of any given quantity to decay
5
NdtdNA λ=minusequiv (111)
The unit is the Becquerel (Bq) with the historical unit being the Curie (Ci) where 1 Ci = 37 x
1010 decays per second and 1 Bq = 1 decay per second In this study the results are given in
terms of activity concentrations which are expressed in units of Bqkg
12 Types of environmental radioactivity
Radioactivity on the earth can be categorized according to three different types namely those
of primordial cosmogenic and anthropogenic nature [Mer97]
bull Primordial radionuclides these have half-lives sufficiently long that they have
existed since the formation of the earth and from the radioactive decay of these
secondary radionuclides are produced (eg 40K 232Th and 238U)
bull Cosmogenic radionuclides primary radiation (protons and other heavier nuclei)
originating from outer space called cosmic rays continuously bombard stable atoms in
the atmosphere and create radionuclides (eg 22Na 7Be and 14C) When cosmic rays
strike the atmosphere they produce a nuclear cascade or a shower of secondary
particles Most of these are eventually stopped before they reach the surface of the
earth except for energetic muons (micro) and neutrons (n) which may penetrate all the
way into the earth
bull Anthropogenic radionuclides these are man-made radionuclides released into the
environment through for example the testing of nuclear weapons nuclear reactor
accidents (eg Chernobyl) and in the radio-isotope manufacturing industry (137Cs 90Sr
and 131I)
6
13 Review of primordial radioactivity
Some of primordial radionuclides that now exist are those that have half-lives comparable to
the age of the Earth (eg 238U 232Th and 40K) Naturally occurring radionuclides can be divided
into those that occur singly and those that are components of chains of radioactive decay
131 Decay series of natural radionuclides
In this study the focus is on gamma-ray emitting daughter nuclei in the decay series of 2238U
232Th and 40K The decay series of 238U and 232Th are characterised more or less by the initial
part dominated by alpha decay and a part dominated by gamma-ray emission This contributes
to the difficulty in the radioactive measurements [Hen01] During a series of radioactive
decays the original radioactive (parent) nucleus N1 decays to another radioactive (daughter)
nucleus until the end of the series where a stable nucleus is formed (206Pb in the case of 238U
series and 208Pb for 232Th) see Figures 12 and 13 The number of parent nuclei N1 decreases
according to the form
dN1 = -λ1N1dt (112)
as discussed in section 113 The number of daughter nuclei increases as a result of the decay
of the parent nuclei and decreases as a result of the decay of its own which leads to
dN2 = (λ1N1 -λ2N2)dt (113)
After a long time equilibrium is reached where the production and the decay rates in the series
are the same [Kra88] so that dN2 = 0 Therefore the activities of all the radionuclides in the
chain must be equal
(λ1N1 = λ2N2 =λ3N3) (114)
7
and this phenomenon is referred to as secular equilibrium This is usually a very accurate
assumption except when daughter nuclei escape for example when the radioactive gas 222Rn
escapes in the decay series of 238U
Sir Humphry Davy discovered the potassium element in 1807 The element is the seventh
most abundant and makes up about 24 by weight of the earths crust Ordinary potassium is
composed of three isotopes one of which is 40K (00118) a radioactive isotope with a half-
life of 128 x 109 years [www01] see Figure 11
t12 = 128 x 109 y
40Ar
40Ca
40K
1032 EC
146 MeV
8928 β-
Figure 11 40K decays by
The above figure shows tw
while on the other hand th
this nuclide is 128 x 109 yea
β+
0001
β+ and electron capture to 40Ar and by β- to 40Ca
o competing decay modes (β+-decay and EC) to the stable 40Ar
ere is 89 chance of decaying by β- to 40Ca The half-life (t12) for
rs
8
uranium (U) 92 25 X105 y
45x109
y
117 m
protactinium (Pa) 91
245 d 75x 104y thorium (Th) 90
actinium (Ac) 89
1600 y radium Ra) 88
francium (Fr) 87
radon (Rn) 86 382 d
astatine (At) 85
140 d 310 m164micros
polonium (Po) 84
50 d 197 m bismuth (Bi) 83
stable 22 y 268 m lead (Pb) 82
3 05 m 132 m thallium (Tl) 81
206 210 214 218 222 226 230 234 238
β α decays
γ-emitting progeny
Figure 12 A Z-A Plot indicating how the 238U nucleus decays the half-lives for the respective nuclides are indicated
9
14 Gy
575 y
305 m
015 s
366 d
556 s
61 h
191y
606 m
106 h606
03 micros
thorium (Th) 90
actinium (Ac) 89
radium (Ra) 88
francium (Fr) 87
radon (Rn) 86
astatine (At) 85
polonium (Po) 84
bismuth (Bi) 83
lead (Pb) 82
thallium (Tl) 81
208 212 216 220 224 228 232
β α decays
γ-emitting progeny
Figure 13 A Z-A Plot indicating how the 232Th nucleus decays the half-lives for the respective nuclides are indicated
10
Most of the radioactivity associated with uranium in nature is due to its progeny that are left
behind in mining and milling The gamma radiation detected by exploration geologists looking
for uranium actually comes from associated elements such as radium (226Ra) and bismuth
(214Bi) which over time have resulted from the radioactive decay of uranium Uranium
constitutes about 2 parts per million (ppm3) of the Earths crust [www02] See Figure 12 for
the decay process associated with this nuclide
In the decay series of for example 238U the parent nucleus 226Ra decays to the daughter 222Rn
by an α decay and 222Rn decays further to 218Po and consequently to 214Pb Since 222Rn is a gas
it might escape the matrix and thus disturbing the secular equilibrium In this event the
activities of the 222Rn progeny will not be the same as their parents and this is an indication of
a disturbance of the secular equilibrium Therefore to obtain secular equilibrium samples are
generally sealed for the radon not to escape This implies that the activity concentrations are
the same for all members of the decay chain When secular equilibrium is reached in the decay
of 238U or 232Th it means that the activity concentrations of these nuclei can be determined by
measuring activity concentration of their γ-emitting progeny The disturbance of secular
equilibrium due to 220Ra (thoron) is less significant due to its short half-life that is 556 seconds
implying that the thoron build up will be negligible
232Th is a naturally occurring radioactive element discovered in 1828 by the Swedish chemist
Jons Jakob Berzelius It is found in small amounts in most rocks and soils where it is about
three times more abundant than uranium Soil commonly contains an average of around 6
parts per million (ppm) of this nuclide The main pathways of exposure are ingestion and
inhalation It was found to be present in the highest concentrations in the pulmonary lymph
nodes and lungs indicating that the principal source of human exposure is inhalation of
suspended soil particles [www02] Figure13 shows the decay process of this nuclide
3 1 ppm U =1225 Bqkg 238U 1ppm Th = 410 Bqkg 232Th 1000ppm = 302 Bqkg 40K [Mer97]
11
14 Literature review natural radionuclide activity concentration in soil
Soil consists of mineral and organic matter water and air arranged in a complicated
physiochemical system that provides the mechanical foothold for plants in addition to
supplying their nutritive requirements [Mer97] The inorganic portion of the surface soils may
fall into a number of textural classes depending on the percentage of sand silt and clay Sand
consists largely of primary minerals such as quartz and has particle size ranging from 60 microm to
about 2 mm Silt consists of particles in the range of 2 to 60 microm while clay particles are
smaller than 2 microm in diameter [Van02a]
The Table 11 shows the values of natural radioactivity in typical soil of volume 1 km times 1 km
and 1 m deep The soil density was assumed to be 16 gcm-3
Nuclide Typical crustal activity concentration (Bqkg)
Mass in 106 m3 Activity in106 m3
238U 25 2800 kg 40 GBq 232Th 40 15200 kg 65 GBq 40K 400 2500 kg 630 GBq 226Ra 48 22 g 80 GBq 222Rn 10 14 microg 95 GBq
Table 11 Amount of naturally occurring radionuclides in typical soil of a volume 1 km times 1 km and 1 m deep [Van02b]
Substantial amounts of radionuclides are found in the earths crust In fact the natural
radioactivity is largely responsible for the fact that the interior of the earth is hot and molten
[Van02b]
The term radiometric fingerprinting describes the identification of mineral species based on
the difference in radionuclide concentrations [Dem97] In soil measurements a correlation
between activity concentrations and grain size has been determined in previous studies While
12
40K activity concentration varies slightly with grain size 238U and 232Th concentrations increased
with an increase in grain size [Van02a] For example Table 12 presents results found in one
case of the difference in activity concentrations for 238U and 232Th which are used to
fingerprint the sediments See also Table 13 for an example of radionuclide fingerprinting with
regard to different minerals
Sediment type 238U-series (Bqkg) 232Th-series (Bqkg)
Sand (gt 63microm) 93 plusmn 09 97 plusmn 09
Mud(lt 63 microm) 462 plusmn 19 456 plusmn 19
Table 12 Characteristic radiometric fingerprints of sand and mud The values are derived from 238U and 232Th activity concentrations of untreated total samples [Van02a]
141 Methods used to determine activity concentrations in soil
There are different methods used in the measurements of activity concentrations in soil These
include laboratory-based measurements using γ-ray methods of analysis and in-situ type of
measurements Examples of these methods will be briefly described in section 143 The focus
in this study is on γ-ray measurements in the laboratory which is based on the principle of
interaction of γ-rays with matter a subject to be discussed in the next section These
interactions need to be well understood in order to clarify results obtained in Chapter 4
13
142 Interaction of gamma rays with matter
There are three primary processes by which γ-rays interact with matter These are photoelectric
absorption Compton scattering and pair production
1421 Photoelectric absorption
Photoelectric absorption takes place when there is an interaction of a γ-ray photon with one of
the bound electrons in an atom as shown in Figure 14 All the energy of the photon is
transferred to the electron The electron is ejected from its shell with a kinetic energy Ee given
by
be EEE minus= γ (115)
where Eγ is the gamma ray energy and Ee the binding energy of the electron in a shell The
atom is left in an excited state with an excess of energy Eb and recovers its equilibrium by de-
exciting and thus redistributing its energy between the remaining electrons in the atom This
may result in the release of further electrons from the atom a process called Auger cascade
[Gil95] A vacancy left by the ejection of the photoelectron may be filled by a higher energy
electron falling into it with the emission of characteristic X-rays (as in Figure 15)
γ-ray
Ee
Eγ Nucleus
Electron Figure 14 A schematic representation of the mechanism of photoelectric absorption where a γ-ray with energy Eγ interacts with a bound electron
14
Nucleus
Kα X-ray
γ-ray Electron
Figure 15 The diagrammatic representation of emission of fluorescent X-rays The cross-section for photoelectric absorption is dependent on the atomic number Z This
varies somewhat depending on the energy of the photon At MeV energies this dependence
goes as Z to the 4th or 5th power The lower the photon energy and the higher the Z of the
material in which this photon interacts the higher the probability of absorption and this
becomes an important consideration when it comes to choosing γ-ray detectors [Leo87]
1422 Compton Scattering
The incident γ-ray interacts with an electron of an absorbing material Part of the γ-ray
energy is then transferred to this electron The energy imparted to the recoil electron is
given by
γγ EEEe minus= (116)
where Eγ is the energy of the incoming photon with E΄γ being the energy of the deflected
photon
15
or
)]cos1[1(
11 20cmE
EEe θγγ minus+
minus= (117)
where m0 is the mass of an electron and c is the speed of light The incoming γ-ray is then
deflected through an angle θ with respect to its original direction Putting different values of θ
into this equation provides a response function with energy Thus with θ =0deg that is scattering
directly forward from the interaction point Ee is found to be 0 and no energy is transferred to
the electron At the other extreme when the gamma ray is backscattered and θ =180deg the term
within curly brackets is still less than 1 so only a proportion of the gamma-ray energy will be
transferred to the recoil electron which gives rise to the Compton edge This corresponds to
maximum energy transferred to recoil electron At intermediate scattering angles the amount
of energy transferred to the electron must be between those two extremes [Gil95] Figure 16
shows Compton scattering
Ee
θ
Eγ
Eγ
Recoil electron
Scattered γ
Figure 16 The diagrammatic illustration of the mechanism of Compton scattering
16
1423 Pair Production
This process as shown on the diagram in Figure 17 is energetically possible if the γ-ray energy
exceeds twice the rest mass of an electron that is 102 MeV The entire energy is converted in
the field of an atom into an electron-positron pair The pair production cross-section does not
become significant until Eγ exceeds several MeV The positron is an anti-electron and after it
slows down and almost comes to rest it will be attracted to an ordinary electron Annihilation
then takes place in which the electron and positron rest masses are converted into two γ-rays
each with energy of 0511 MeV These annihilation γ -rays are emitted in opposite directions to
conserve momentum and they may in turn interact with the absorbing medium by either
photoelectric absorption or Compton scattering [Lil01]
In this study the focus is on γ-rays of energies less than 3 MeV thus pair production does not
play a major role The cross sections for this process are higher at energies above 3 MeV as is
confirmed in Figure 18
posit
electronIncident γ-ray
elect511 keV
Figure 17 The diagrammatic illustra
E
Eγ
ron
ron 511 keV
tion of pair production
17
1424 Attenuation Coefficients
The attenuation coefficient is defined as a measure of the reduction in the gamma-ray intensity
at a particular energy caused by an absorber [Gil95] Photoelectric interactions are dominant at
low energy Compton scattering at mid-energy ranges while pair production is dominant at
high energies In the low-energy region discontinuities in the photoelectric curve appear at
gamma-ray energies which correspond to the binding energies of electrons in the various
shells of the absorber atom The edge lying highest in energy corresponds to the K shell
electron For gamma-ray energies slightly above the edge the photon energy is just sufficient
to undergo a photoelectric interaction in which the K electron is ejected from the atom For
gamma rays below the edge this process is no longer energetically possible and therefore the
interaction probability drops abruptly Similar absorption edges occur at lower energies for the
L M electron shells of the atom [Kno79]
The total cross section σtot for the photon atom interaction may be written as
cpppetot σσσσ ++= (118)
where σpe σpp and σc are respectively the cross-sections for photoelectric absorption pair
production and Compton scattering [Cre87]
Present tabulations of the mass attenuation coefficient microρ rely on theoretical values for the
total cross section per atom which is related to microρ according to
)][(22
Aguatomcm
gcm
tot
=
σρmicro (119)
u (= 167 x 10-24 g) is the atomic mass unit and A is the relative atomic mass of the
target element
18
Figure 18 The linear attenuation coefficient of germanium and its component parts on a log-log scale [Gil95]
On multiplying the mass attenuation coefficient microρ by the density ρ of the material the result
will be the linear attenuation coefficient micro This phenomenon is an important part of this
study and will therefore be discussed in more detail in the Appendix D under the discussion
on self-absorption effects
143 Examples of gamma-ray spectrometry systems
bull In-situ
The successful demonstration of the in-situ γ-ray spectroscopy for the rapid and accurate
assessment of radionuclides in the environment has been witnessed over time This allows for
the measurement of γ-ray intensities in the soil without any soil sampling and treatment
Although soil sampling and subsequent laboratory analysis is still needed for confirmation of
results the number of samples required can be reduced considerably
19
The accuracy of in-situ measurements in determining radionuclide activity concentrations in
soil relies on two primary elements the detector calibration and source geometry of the site
The field calibration can be expressed in terms of measured full absorption peak count rate N
(the subscript o represents the response at the normal incidence and the subscript f
represents the integrated response over all angles) the fluence rate Φ and the activity
concentration in the medium A [Mil94] The ratio of these quantities is then expressed as
A
NNN
AN O
O
ff Φbull
Φbull= (120)
where NfA is the total absorption peak count rate (cps) at some energy E for a particular
radionuclide per unit concentration of that radionuclide in soil NfNO is the angular
correction factor for radionuclide distribution in the soil NOΦ is the full energy peak count
rate to flux density for parallel beam of photons at energy E that is normal to the detector face
ΦA is the photon flux density from un-scattered photons arriving at the detector per unit
radionuclide activity for a given radionuclide distribution in the soil
The source geometry is evaluated as a volume source at a fixed height of one metre above the
ground The source geometry is primarily controlled by the depth profile of the radionuclide
being measured
A typical example of a field γ-ray measurement is a conventional portable coaxial HPGe
detector with a 12 relative efficiency and a resolution of 19 keV (relative to the 1332 keV for 60Co) A tripod supports the detector with the front part being 1 meter above the ground The
dewar has a capacity of 70 liters of liquid nitrogen (LN2) and holding time of five days The
detector is orientated in a downward facing way Detector calibrations for field γ-ray
spectrometry are performed by calculations and by using point like γ-ray sources [Fuumll99]
20
bull Laboratory based gamma ray spectroscopy
γ-radiation is a penetrating form of radiation and therefore can be used for non-destructive
measurements of samples of any form and geometry as long as standards of the same form are
available and are counted in the same geometry to calibrate the detector Various detector
types are used to measure γ-rays The one used in this study is the HPGe which has the
advantage of high-energy resolution
Most γ-ray spectrometry systems are calibrated with commercially mixed standards ideally the
matrix and geometric form of standards needs to match that of sample as closely as possible
However this is often difficult because one does not for example know the composition of the
sample to be analysed There is commercially available software for the analysis of the γ-ray
spectra Adjustments to the calibration for the corrections of density and effects such as self-
absorption need to be done This study utilises this method of analysis
15 The motivation for this study
The research interests in the Environmental Radiation Laboratory (ERL) at iThemba LABS
(South Africa) are as shown in the diagram below
Applied Radiometry
Environmental ApplicationsAgricultureMining explorations
Figure 19 Application of Radiometric methods in different fields
21
A connection between radiometry and minerals has been established and a method called
radiometric fingerprinting was the result [Dem98] In principle characterisation of samples is
based on the minerals percent composition of natural radionuclides such as 238U 232Th and 40K
by detection of the gamma rays from such minerals Samples are collected from the area of
surveillance and brought to the laboratory for analysis
Table 13 shows the results of a study on radiometric fingerprint of minerals associated with
heavy minerals deposits conducted in South Africa [Dem98]
Mineral 40K 214Bi 232Th
Quartz 116 (8) 2 (2) lt 9
Siltclay minerals 218 (10) 494 (11) 127 (3)
Ferricrete 95 (7) 130 (3) 160 (4)
Magnetite lt 10 120 (120) 200 (200)
Ilmenite 28 (05) 18 (2) 27 (9)
Rutile 13 (3) 460 (70) lt 60
Zircon lt 18 3280 (120) 444 (16)
Monazite 1600 (600) 42000 (2000) 181000 (2000)
Table 13 Radiometric fingerprint given as activity concentrations (Bqkg-1) associated with heavy minerals in South Africa [Dem98] Uncertainties are given in brackets
22
The table shows that the values of natural radioactivity concentrations can be used to
characterise minerals according to grain size and composition
The Environmental Radioactivity Laboratory (ERL) has embarked on measurement of natural
and anthropogenic radionuclides in soil and liquid samples For this purpose a high purity
germanium detector (HPGe) housed in a lead castle is being used for laboratory-based
measurements while a MEDUSA (Multi Element Detector System for Underwater Sediment
Activity) scintillation-based detector is being used for in-situ measurements [Hen01] This
study will discuss a new method called the Full Spectrum Analysis (FSA) method in addition to
the conventional Window Analysis (WA) method to measure activity concentrations in soil
samples (see the next section) in the laboratory
16 The aim and scope of this study
Traditionally activity concentrations in soil samples are determined using what is called a
Windows Analysis (WA) procedure This involves analysing the spectrum measured (normally
in the laboratory) using windows or regions of interest (ROI) set around prominent γ-ray
peaks associated with the decay of 238U 232Th and 40K The windows are used to determine the
net counts (that is after an appropriate background subtraction) in the peak of interest By
using these counts with the branching ratio (associated with a particular γ-decay path)
measurement live time sample mass and full energy peak detection efficiency it is possible to
calculate the activity concentrations
A new technique for measuring primordial radionuclide activity concentration in soil samples
with HPGe is introduced in this study This technique called Full Spectrum Analysis (FSA)
uses the full spectral shape and standard spectra to calculate the activity concentrations of
238U 232Th and 40K present in a geological matrix [Hen01]
The FSA technique was first used in conjunction with the MEDUSA detector by the KVI
Nuclear Geophysics Division [Hen01] The MEDUSA detector which detects γ-radiation by
23
means of a scintillator (BGO or CsI) is used to perform in-situ measurements of natural
activity concentrations on seabeds [Ven00] and on land [Hen01]
FSA involves the fitting of a measured γ-ray spectrum (~200 keV to 3 MeV) by means of a
linear combination of the three so-called standard spectra and a background spectrum Each
standard spectrum corresponds to the response of the detector in a particular geometry
(normally that of a flat bed) for a sample containing 1 Bqkg of a particular nuclide (238U 232Th
and 40K) These standard spectra are normally obtained via measurement or by means of
Monte Carlo simulations The measured spectrum S is considered to be a superposition of
weighted standard spectra where the factors Ck CU and CTh are the activity concentrations of
each nuclide in the sample [Dem97] Mathematically it can be expressed as
)()()()()( iSiSCiSCiSCiS BThThUUKK +++= (121)
The spectrum deconvolution was applied and the coefficients CK CU and CTh were determined
using the χ2 minimisation technique
In this study the results from FSA and WA of HPGe spectra of a variety of sediment samples
are critically compared after which the advantages and disadvantages of each method are
established
17 Thesis outline
The next chapter will talk about the experimental techniques The HPGe detector system used
will be described in detail in chapter 2 while associated experimental work and data analysis
are discussed in chapter 3 The results will be presented and discussed in chapter 4 Finally a
summary and conclusion will be given in chapter 5
24
Chapter 2
Experimental Methods
Various types of detector differ in their operating characteristics but all are based on the same
fundamental principle the transfer of part or all the radiation energy to the detector mass
where it is converted into an electrical pulse The form in which the converted energy appears
depends on the detector and its design [Leo87] In this study a high purity germanium (HPGe)
detector is used The operation of this detector involves the following first the photon energy
is completely or partly converted into kinetic energy of electrons (and positrons) by
photoelectric absorption Compton scattering or pair production second the production of
electron-hole pairs third the collection and measurement of charge carriers
The various components of the HPGe detector used will be discussed in this chapter The
process followed in executing measurements will also be described The performance
characteristics of the detector will also be presented
21 The HPGe gamma-ray detection facility
The facility is used to measure natural and anthropogenic activity concentrations in soil and
liquid samples (though in this study the focus is on soil samples) The gamma ray
spectroscopy is mainly used for the analysis of natural gamma rays from potassium and the
progenies of uranium and thorium It can also be used to measure artificial radioactivity such
as the activities from cesium strontium etc The available equipment in the laboratory
includes an HPGe detector lead (Pb) shielding Marinelli beakers (for sample holding) PC-
based data acquisition system digital weighing balance and standards
25
211 Physical layout
Figure 21 shows experimental set up used in the present study (the picture was taken in the
Environmental Radiation Laboratory (ERL))
Lead Castle (detector inside) Amplifier
NIM crate
Dewar MCA card inside Oscilloscope DesktopHose (LN2
filling)
Figure 21 The picture showing the setup of the Environmental Radioactivity Laboratory at iThemba LABS
The lead castle which opens from the top shields the detector and the dewar underneath is
for holding liquid nitrogen (LN2) The oscilloscope is used to set and monitor the pulse
properties while the NIM crate carries the amplifier After amplification the signal is processed
in the MCA card in the PC and then displayed to the desktop
All radioactive sources are kept in the storeroom far away from the set up (approximately 5
meters) in order to avoid the contribution of such sources to the background during counting
26
The laboratory has an air conditioning facility that ensures the maintenance of the normal
temperatures around 18 degC
bull Lead Castle
Lead is the material employed in the shielding of the detector used in this study It makes a
good shielding material due to its high density and large atomic number A standard 25
(relative efficiency) detector measurement in a 10 cm thick lead shield will reduce the
environmental background by a factor of 1000 thus enabling one to measure
301000 asymp times weaker sources with the same statistical accuracy [Ver92]
In this study a 10 cm thick lead castle was used The inside of the castle was lined with a
copper layer of 2 mm thickness The copper layer is there to attenuate the X-rays from the lead
(see the Figures 22 and 23) A typical background spectrum measured with the ERL HPGe is
shown in Appendix C
Figure 22 The picture shows the lead castle supported by a metallic frame surrounding the HPGe detector
27
The black hose shown in Figure 22 is for the filling of the detector dewar with liquid nitrogen
A B
Figure 23 The open top view showing the detector (A) and a Marinelli beaker on top of the detector (B)
212 HPGe technical specifications
The detector used is a closed end-coaxial Canberra p-type (model GC4520) with a relative
efficiency of 45 The germanium crystal is located inside the lead shield The vertical dipstick
cryostat (model 7500SL) which is cooled by liquid nitrogen is contained inside a dewar (see
Figure 27) The detector crystal has a diameter of 625 mm and a length of 595 mm
213 Electronics
The electronic system for this semiconductor detector (HPGe) is shown schematically in
Figure 24 The system consists of a detector bias supply (SILENA model 7716) preamplifier
(model 2002CSL) amplifier (model ORTEC 572) multi channel analyser (OXFord-Win) and
a desktop computer
28
MCA amplifier
desktop
preamp
Bias supply
detector
Figure 24 Schematic diagram illustrating the electronic setup used to acquire the HPGe
data
bull Detector
The detector is a cylinder of germanium with an n-type contact on the outer surface and the
p-type contact on the surface of an axial well see Figure 25 The n and p contacts are diffused
lithium and implanted boron respectively [USR98] The germanium detector like other
semiconducter detectors is a large reverse biased p-n junction diode The germanium has a net
impurity level of about 1010 atomscm3 so that with the moderate reverse bias the entire
volume between the electrodes is depleted and the electric field extends across this region
[USR98]
29
Figure 25 Coaxial Ge Detector Cross Section [USR98]
The radiation energy is transferred to the detector crystal by an interaction process (photo
electric interaction) which then creates electron-hole pairs
h+
Valence band
Conduction bande-
Eg
Figure 26 A diagram showing a typical semiconducter with band gap Eg
The mobile electrons in the conduction band were promoted under an influence of an electric
field from the valence band the holes in the valence band are also mobile but the mobility of
the latter is in the opposite direction to the former This movement of electron hole pairs (e-
h+) in a semiconducter constitutes a current If these arrive at their respective electrodes the
30
result is a current proportional to the energy deposited in the crystal or a pulse [Lil01] This is a
small signal requiring amplification
The energy required to create an electron-hole pair across the Ge band gap (Eg) is
approximately 3 eV thus an incident gamma ray with an energy of several hundred keV
produces a large number of such pairs leading to good resolution and low statistical
flactuations These are desireable properties of a detector HPGe detectors are operated at
temperatures of around 77 K in order to reduce noise from electrons which may be thermally
excited across the small band gap at standard temperatures [Lil01] There are weekly fillings of
the ERL HPGe liquid nitrogen dewar (capacity of about 20 liters) to provide for such low
temperatures
The following is a cross sectional diagram of a germanium detector with a dewar
Figure 27 A typical germanium detector cryostat and liquid nitrogen
reservoir(dewar)[Gil95]
31
bull Detector bias
Radiation detectors require the application of an external high voltage for their proper
operation This voltage is normally referred to as detector bias and the high voltage supplies
used for this task are often called detector-bias supplies [Leo87] The SILENA model 7716
detector bias supply was used in this study A bias voltage of approximately 35 kV was
supplied and as photon interaction takes place within the depleted region charge carriers are
swept by the electric field to their collecting electrodes The specified depletion voltage for the
HPGe used is in fact 3 kV
bull Preamplifiers
The main function of the preamplifier is to amplify the weak signal and send it through to the
cable that connects the preamps with the rest of the equipment The preamps are mounted as
close as possible to the detector to minimise the length of the cable since the input signal is
very small The longer the length of the cable the more the capacitance and that reduces the
signal-to-noise ratio [Leo87] Therefore the manufacturing of the built-in preamplifiers
minimises this effect Furthermore when the detector system is cooled the preamplifier also
indirectly gets cooled thus reducing the chances for thermal excitation of charge which could
bring about leakage current4 A charge sensitive preamplifier will convert the charge into a
voltage pulse proportional to the energy deposited in the detector crystal The preamplifier
used in this study is of the model 2002CSL
bull Amplifier
The amplifier (model ORTEC 572) then further amplifies the pulse from the preamplifier to a
bigger size The pulse needs to be shaped to a more convenient form therefore the amplifiers
4 The unwanted current leaking between two electrodes under voltage
32
frequency response is set by a shaping time constant τ which is 6 micros for the amplifier
[USR98] Should a second signal arrive within the given period τ it will ride on the tail of the
first and its amplitude will be increased The energy information will therefore be distorted
resulting in a phenomenon called pile-up which is when pulses are too close together to be
separated [Leo87] This is not a problem in this study since the detector system dead time was
always less 1 Pulse shaping can also be used to optimise the signal to noise ratio
bull Multi Channel Analyser (MCA)
The operation of the multi channel analyser is based on the principle of converting an analog
signal which is basically the pulse amplitude into an equivalent digital number usually referred
to as a channel After this is done the digital information will be stored in the memory to be
displayed on the monitor This activity is in principle carried out by the Analog to Digital
Converter (ADC) [Kno00] The pulses are collected sorted according to pulse height in the
ADC and a γ-ray spectrum is generated Therefore the performance of the MCA is primarily
dependent on the ADC The results discussed in this thesis were measured using an MCA card
(OXFord-Win) in a PC using the OXWIN software program The MCA diagram is shown
below
plotter printer disc Other output devices
Monitor
MemoryControl logicA D C
Figure 28 Basic architecture of a MCA
33
214 Figure of merit
To keep track of the performance and reliability of the detector it is imperative to keep on
checking our results based on the detector specifications This can be achieved by doing the
energy calibration checking the Full Width at Half Maximum (FWHM) and determining the
peak to Compton ratio The background measurements were done in each case to ensure
derivation of proper concentrations These concepts are discussed in this section
bull Energy calibration
Prior to acquisition of the data energy calibration has to be performed as part of the setting up
procedure Various techniques of determining the energy at a particular channel are
implemented as this is a requirement by most nuclear applications In some MCAs a simple
two-point energy calibration is used to determine both the offset and slope by the equation
BchAE += )( ( 21)
where E is the energy and ch is the channel number and A and B are constants thus the
energy as a function of channel number can be directly read out The MCA systems in use
allows the user to choose between first-order (linear) or second order (quadratic) equations
that use a least square fit to data points In this study a second order equation was used
because the detection and amplification are not always exactly linear and this can be written as
follows
(22) CchBchAE ++= )()( 2
34
Any source whose peaks are known can be used to do the energy calibration In our study
sources such as a mixed liquid source was used (152Eu 60Co and 137Cs mixture) 232Th and 238U
sources were also often used The following is a particular case in which a mixture of 40K 232Th
and 238U was used as a source The energies of prominent γ-ray lines are presented in Table 21
below
Decay series
Decaying nucleus Energy (keV) FWHM (keV)
238U 226Ra235U 1861 151 232Th 212Pb 2386 154 238U 214Pb 2951 157 232Th 228Ac 3384 160 238U 214Pb 3520 160 232Th 208Tl 5830 173 238U 214Bi 6093 174 232Th 212Pb 7273 181 232Th 228Ac 9112 191 238U 214Bi 11203 203 40K 40K 14608 221 238U 214Bi 17645 238 238U 214Bi 22041 262 232Th 208Tl 26144 285
Table 21 γ-ray energies with associated Full Width at Half Maxima (FWHM) for γ-ray lines associated with decay of 40K and the series of 238U 232Th the energies are taken from [EML97]
The objective here is to derive a relationship between peak position in the spectrum and the
corresponding gamma-ray energy The mixture of three sources with well-known energies was
made to ensure that the calibration covers the entire energy range over which the spectrometer
is to be used The data on energies were taken from [EML97]
Figure 29 shows the spectrum for the mixture of materials used
35
214Pb
0
02
04
06
08
1
50 100 150 200 250 300 350 400 450 500
0
02
04
500 600 700 800 900 1000
0
02
04
06
08
1
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0
005
01
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
005
01
015
02
2000 2100 2200 2300 2400 2500 2600
226Ra 212Pb 214Pb228Ac
208Tl 214Bi
214Bi
40K 214Bi
228Ac
208Tl
Cou
nts
seco
nd
Figure 29 The energy calibration spectrum showing the lines (labelled) used to performcalibrations a mixture of 40K 232Th and 238U was used as a source
Energy keV
36
energ
y
The spectrum has to be measured for long enough in order to determine the peak properties
with sufficient accuracy for the peaks to be used for the calibration The calibration process
involves marking the peaks to be used and their true energy see section 31 for the region of
interest (ROI) discussion The set ROI is used by the Oxwin software to calculate amongst
others the peak centroid
The relationship between energy and channel number is shown in Figure 210
R2 = 1
0
500
1000
1500
2000
2500
3000
0 1000 2000 3000 4000 5000
Channel
Ener
gy (k
eV)
A = 2 x 10-8
B = 06427 C = -02174
Figure 210 A typical fit to data points used to energy calibrate the HPGe detector
system The energies used are also given in Table 21
37
bull Energy resolution
The energy resolution of the HPGe detector system as a function of γ-ray energy was
calculated after the energy calibration The energy resolution of the detector (at a particular γ-
ray energy) is conventionally defined as the full width-to-half maximum (FWHM) of the peak
divided by the peak energy Therefore the energy resolution which is a dimensionless fraction
is conventionally expressed in percentages Small percentage resolution of a peak implies good
resolution [Kno79]
In principle one should be able to resolve peaks at two energies which are separated by more
than one value of the detector FWHM at that energy
In order to parameterize the dependence of FWHM on γ-ray energy the FWHM data for the
lines given in Table 21 were fitted using a linear function The fit to the data are shown in
Figure 211 The results for the FWHM calibration shown in the Table 21 were obtained
from the following equation
BAEFWHM += (23) where A and B are constants deduced by the OXWIN software program and E is the γ-ray
energy The graph in Figure 211 was then plotted from these results
38
R2 = 1
00
05
10
15
20
25
30
0 500 1000 1500 2000 2500 3000Energy (keV)
FWH
M (k
eV)
B = 00006 C =1409
Figure 211 A Full Width at Half Maximum calibration graph with a line of best fit
According to the specifications of the detector the resolution should be 2 keV for the 1332
keV line for 60Co On interpolation the value of 22 keV was found for this value in this work
implying a deviation by 9 from the specified value
bull Peak to Compton Ratio
The Peak-to-Compton ratio is an important quantity defined as the ratio of the counts in the
channel corresponding to the highest number of counts per channel in the photo-peak (photo-
peak centroid) to the counts in the typical channel of the Compton continuum This typical
channel is defined as the region of interest between 1040-1094 keV for a 60Co source [Kno00]
The Compton continuum results from the Compton scattering in which the gamma rays
entering the detector will not deposit their full energy on interaction with matter This partial
energy event appears in the spectrum as an event below the full energy peak in the Compton
continuum The Peak-to-Compton ratio ranges from 30 to 60 for coaxial germanium detectors
when using the 1332 keV line associated with the 60Co decay [Kno00]
39
00001
0001
001
01
1
10
100
0 150 300 450 600 750 900 1050 1200 1350
Energy (keV)
Cou
nts
seco
nd (l
og s
cale
) 1332KeV1173KeV A
ROI
Figure 212 A spectrum of 60Co for peak to Compton ratio determination
A region of interest ROI (10401096) keV was established along the Compton continuum
and then the ratio of the number of counts in the photo peak to the continuum was
determined The Table 22 shows the data required for such a measurement
A ROI C G
30394 511 87 44482
Table 22 A table of counts in the chosen region of interest with number of channels along
the 60Co Compton continuum and in the channel corresponding to the highest number of
counts in the full energy peak A = Number of counts in the channel corresponding to the
highest number of counts per channel in the full energy peak ROI = Counts per channel in
the region of interest C = Number of channels in the region of interest G = Gross counts in
the region of interest
Counts per channel in the region of interest ROI = G C = 511 = Gross counts divided by
the number of channels within the region Therefore peak to Compton ratio = AROI = 59 1
40
This value is very close to the one specified by the manufacturer which is 581 [USR98] The
higher the value the more efficient the detector is and thus the value should preferably be
high
22 Sample Selection Preparation and Measurement
The types of samples studied were mainly soil taken from mine dumps in particular from
Kloof mine in Westonaria south west of Johannesburg (RSA) and beach sand from the west
coast of South Africa The samples were packed in plastic bags from the areas of surveillance
and brought to the laboratory at iThemba LABS At the laboratory the soil samples were put
in an oven (Labotech) and set to 105ordmC to allow for drying overnight After drying the samples
were crushed and sieved with a mesh having a hole diameter of 2 mm in order to remove
organic materials stones and lumps The information about treatment of samples can be
obtained from the general guide [ISO03] The samples were weighed on a digital weighing
balance (Sartoslashrius model BP2100S) with a precision of plusmn 001g and after sealing with silicon
sealant in Marinelli beakers the samples were allowed to stand for several days (about 21 days)
for secular equilibrium to be reached The following is a table of sample information
Sample ID Mass(kg) Livetime (s) Volume (ml) (Estimated)
Density (gcm3)
Sealed YesNo
Kloof sample 6B
121477 35924 1000 121 Yes
Westonaria Sand 17A
126737 74357 800 158 Yes
West Coast sand (3)
142652 28796 900 159 Yes
Brick Clay (6)
115201 28776 860 134 Yes
Thorium+stearic acid 5
067734 28740 1000 0677 Yes
Table 23 The table of the samples collected at different sites
41
bull Marinelli beaker
Environmental samples of low-level radioactivity are often measured in Marinelli beakers
specially designed to provide good detection sensitivity Full Energy Peak Efficiency (FEP)
variations are observed in this geometry for different sample types [Sim92] this is due to self-
attenuation effects a concept that is discussed later with results (see also Appendix D) See
Figure 213 and D1 for Marinelli beaker photo and a drawing of its dimensions respectively
Figure 213 Photo of Marinelli beaker and the design of its geometry The bottom part shows its re-entrant hole [Ame00]
In the Environmental Radiation Laboratory (ERL) at iThemba LABS the 1-liter Marinelli
beaker (Model VZ-1525) made of polypropylene material manufactured by Amersham
[Ame00] was used The precision with which the activity can be determined with this Marinelli
geometry is limited by the effect of self-absorption for which correction must be made
[Dry89] Self-absorptions effects were verified through the use of the Sima model [Sim92]
42
This was done on the basis of the difference in the samples density and volume in this study
The results for this will follow in chapter 4
In window analysis if the standard source has the same physical dimensions density chemical
composition and radionuclides present as in the sample the radioactivity of the sample is
simply determined by comparison of the count rates in the corresponding full energy peak of
the measured pulse height gamma ray spectrum [Par95]
23 Reference sources
The two reference sources IAEA-RGU-1 and IAEA-RGTh-1 prepared by the Canada Center
for Minerals and Energy Technology on behalf of the International Atomic Energy Agency
were used in this work [RL148] The ERL Laboratory at iThemba LABS bought the 40K (KCl
powder) reference
bull WA
The reference source used in this case was KCl of 99 purity this was used specifically for the
efficiency calibration The advantage of this standard source is the fact that it is easier to
handle and is not that hazardous The method of calibrating detector efficiency using this
standard is described in the next chapter The activity of this is given in the Table 24 below
This standard was also used in the FSA method of analysis
bull FSA
The information of the reference sources used in the FSA method of analysis is tabulated in
Table 24 The full details regarding the preparation of these are given by [RL148] except for
the 40K reference source in which KCl was used instead of K2SO4 as used by the IAEA
(reference manual [RL148] ) for example
43
Nuclide Used in which
method
Source Supplied in what form
Mass (kg)
Volume (ml)
Density (gcm3)
Activity Concentration
(Bqkg) 238U FSA IAEA (RGU) 04873 400 12 4940
232Th FSA IAEA (RGTh) 05157 400 13 3250
40K FSAWA KCl (99purity) 12721 1000 13 16258
Table 24 The table shows the data of reference materials used
The energy calibration was done independently for each source and the sample to be
measured It is seen in Table 24 that the volumes of the reference sources for uranium and
thorium differ significantly from those for the sample The effects of this on the results
presented below are discussed in chapter 4
24 Data acquisition
Each time a measurement is done energy calibration has to be done as part of the setting up
procedure before any data acquisition The counting time was preset on the Oxford-Oxwin
software used
Information regarding the spectra acquired with reference sources samples and background
(Marinelli filled with tap water) are given in Tables 25 and 26
44
Source type Measurement date
Elapsed Real time (s)
Elapsed Live time (s)
KCl 150802 16516 16436
RGU 40602 16200 15996
RGTh 60502 16200 16026
Table 25 Spectral information on reference sources (see Table 24 for information on Sources)
Long Measurement Short Measurement Sample
Code Elapsed
real time(s)
Elapsed live
time(s)
Elapsed real time(s)
Elapsed live
time(s) Kloof 6B
36000 35925 3600 3594
Westonaria 17A
74500 74357 4053 4046
West Coast Sand D3
28800 28796 3600 3599
Brick clay D6
28800 28776 3600 3594
Thorium+ stearic acid 5
28800 28740
Marinelli filled with tap water (Background)
116322 116312
Table 26 Spectral information on Samples and background (see Table 23 for more information on the samples)
45
Chapter 3
Data Analysis
In this chapter the two methods used in this work for determining the activity concentration
of 238U 232Th and 40K in sediments namely the Window Analysis (WA) and Full Spectrum
Analysis methods (FSA) are described
31 Window Analysis
In the Window Analysis method the activity concentrations A (Bqkg) in sediments are
calculated using the equation
)(EtiB
iN
LMrC
kgBqAε
prime= (31)
where is the net counts in the ith γ-ray peak associated with the nucleus of interest (primeiNC 238U
232Th or 40K) Lt is the live time associated with the sample spectrum M is the mass of the
sample εE the full energy peak detection efficiency at a particular energy E and rBi the
branching ratio of the energy line used (see Table 31 for branching ratios used in this study)
primeiNC is obtained from an analysis of the peak of interest using a region of interest (ROI) or
window set around the peak hence the term Window Analysis An example of a window set is
shown in Figure 31 where the 1461 keV line (40K) is focused on In this case the window starts
at an energy K (~1455 keV) and ends at an energy Q (~14665 keV) The region below line P
is due to the Compton continuum
In this study CNi was determined using the Oxwin MCA software The software calculates the
gross counts Cg in the window of interest (starting at channel s and ending at channel e)
according to the equation 32
46
(32) sum=
=
=ei
siig CC
where Ci is the counts in channel i The counts in the continuum are calculated using the
equation
(33) sum=
=
=ei
siCic CC
where CCi is the continuum counts in channel i
The software can therefore calculate the net counts CNi in a particular photopeak from
cgNi CCC minus= (34)
Even when a sample is not being counted by the HPGe counts are registered in the spectrum
due to room background (see Figure 32 for a sample and background spectra) The
contribution to CNi from room background has to therefore be subtracted A room
background spectrum was measured by using a Marinelli beaker filled with a 1 liter of tap
water This spectrum is shown in Appendix C The windows used in the analysis of the sample
spectra were used to determine Cbi the net background counts in the peaks of interest
The background subtracted net counts CNi in the ith peak was calculated using the expression
ibB
CiNiN C
LL
C
minus=primeC (35)
where LC and LB are the detector live times during the measurement of sample and room
background respectively
47
0001
001
01
1
1450 1452 1454 1456 1458 1460 1462 1464 1466 1468 1470
Energy (keV)
Cou
nt ra
te (l
og s
cale
)
K CN
P Continuum
Figure 31 A graphical representation of the region of interest with window setting around the 40K peak
As can be seen from equation 31 the full-energy peak (also termed photo peak) detection
efficiency as a function of γ-ray energy is needed in order to calculate activity concentrations
and is discussed in the next section
48
000001
00001
0001
001
01
1
10
50 550 1050 1550 2050 2550
Energy (keV)
C
ount
sse
cond
(log
sca
le) Background
Sample 6
Figure 32 A typical representation of sample (number 6b Kloof sample) spectrum with
superimposed background spectrum (measured using a Marinelli filled with 1litre of water) on a log scale
311 Determination of γ-ray detection efficiency
The method for determining the efficiency curve used in this work involves two steps The
first step involves the measurement of the relative detection efficiency using 238U and 232Th
lines as observed in the sample spectrum measured after secular equilibrium is achieved The
second step entails placing the curve on an absolute scale by using the 1461 keV (40K) line This
is done by calculating the absolute efficiency at 1461 keV for a Marinelli beaker filled to 1 litre
with KCl This is similar to the technique described in reference [Fel92]
49
One advantage of this approach is that the self-absorption effects are automatically taken into
consideration [Fel92] The other advantage is the fact that the KCl source is more convenient
in the sense that it is easier to handle from a safety point of view
It is assumed that the two decay chains are in secular equilibrium
The relative efficiency data were fit with a function of the form
b
EEa
=
0
ε (36)
where ε is the efficiency E is the γ-ray energy in keV (E0 = 1) with a and b being fit
parameters [Dry89]
On taking the logarithm of equation (36) one obtains
ln (37) 0
lnlnEEba +=ε
50
A flow-chart illustrating the steps used in more detail is given in Figure 33
1Calculation of relative
efficiencies
Curve fitting (ε = aEb)
2
Determination of Activity Concentration for KCl
Reference source 3
Determination of efficiencyat 1461 keV (using KCl)
4
Determination of the ratio
between 2 amp 4 5
Multiply the value in 5 by 2 6
Construction of absolute
efficiency
7
Figure 33 A flow chart showing the steps followed in constructing the absolute efficiency curve
51
The 238U and 232Th γ-ray lines used to construct the relative efficiency curves along with other
properties are given in Table 31 Generally the lines used have large branching ratios
However some lines such as the 609 keV and 583 keV lines associated with the decay of 214Bi
and 208Tl respectively were omitted since they are prone to coincidence summing [Gar01]
Furthermore lines overlapping with others were not used with the only exception being the
186 keV line (doublet due to lines 238U235U) and the 967 keV line due to 228Ac The γ-ray lines
from the radionuclide lines used in the efficiency calibration are shown on the Table 31 (and
Figure 33)
Nuclide
Energy (keV)
Branching ratios
238U Series
226Ra 1861 005789 214Pb 2951 0185 214Pb 3520 0358 214Bi 7684 005 214Bi 9340 0032 214Bi 11203 015 214Bi 12381 0059 214Bi 13777 004 214Bi 17645 0159 214Bi 22041 005
232Th Series 228Ac 3384 0124 212Bi 7273 0067 228Ac 7948 0046 208Tl 8603 0043 228Ac 9112 029 228Ac 9668 0232
40K Series
40K 14608 01066
Table 31 The gamma ray lines used and their associated branching ratios the branching
ratios are taken from [EML97] branching ratio for 226Ra was corrected for the fact that it
is doublet with a 185 keV peak due to 235U
52
0
10000
20000
30000
40000
50000
60000
50 100 150 200 250 300 350 400 450 500
0
500
1000
1500
2000
2500
3000
3500
4000
500 550 600 650 700 750 800 850 900 950 1000
0
1000
2000
3000
4000
5000
6000
7000
8000
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
0500
100015002000250030003500400045005000
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
0
200
400
600
800
1000
1200
2000 2100 2200 2300 2400 2500 2600
214Pb
214Pb
226Ra235U
214Bi
214Bi
228Ac
40K
214Bi214Bi
214Bi
Cou
nts
chan
nel
228Ac
208Tl
214Bi 228Ac212Bi
Energy (keV)
Figure 34 The spectrum of Kloof sand sample showing the lines used (see Table 31) duringdetection efficiency calibration
53
From the uranium series fit parameters (a and b) were determined using equation 37 These
parameters are used to determine the factor that is needed to join the thorium content to the
uranium content line using the equation 36 Calculations of the relative efficiencies for all the
lines including the 1461 keV were done and at this point the relative efficiency curve is
determined The fit parameters generated for a particular sample (Kloof sample) are presented
in the graph below
-16-14-12
-1-08-06-04-02
0020406
0 2 4 6 8
ln(E)
ln(r
elat
ive
effic
ienc
y)
10
238U
a = 41852 b = -07241
Figure 35 Fit of relative efficiency (with parameters a amp b) as determined from lines associated with the decay of 238U (for a Kloof sample number 6) Values were normalised relative to the 352 keV line
The Figures 36 and 37 depict the relative efficiency curve from thorium points and the curve
from a combination of both 238U and 232Th points respectively From the relative efficiency
curve in Figure 37 a normalization factor (step five from the flow chart in Figure 33) is
needed to multiply all the values corresponding to the points in Figure 38 to obtain the
absolute efficiency curve The absolute efficiency curve for this sample is shown in Figure 39
54
The KCl sample was used as a standard source and the absolute efficiency calibration was
determined using this sample The activity of 40K was calculated as follows
From equation (115) Nλ=Α
where A is the activity with λ as the decay constant and N the number of 40K nuclei in KCl In
order to obtain N one needs to find the number of moles in the KCl and therefore multiply
that with the Avagadros number NA = 6022 x 1023 This brings us to the number of moles n
as in the equation (38)
Mmn = (38)
with m as the mass of the KCl (1000 g) source and M the molar mass (74551 gmol)
Since (39) aNnN A timestimes=
with a being the abundance and that
21
2lnt
=λ from equation (114)
where t12 the half life for 40K is 1277 x 109y
Multiplying N by the activity constant λ and the abundance a of 40K in KCl which is equals to
117 x 10-4 gives the activity 16256 Bqkg
The KCl spectrum measured is shown in Figure 315 (section 321)
The absolute full-energy peak detection efficiency was then calculated using the expression in
equation 310
55
ALMr
C
tB 1461
1461 =ε (310)
where C1461 is the nett counts in the 1461 keV line (after background subtraction) and the
other symbols are as determined from equation 31 ε was found to be 000940 plusmn 000007 The
uncertainty quoted is that associated with counting statistics
This efficiency divided by the relative efficiency at 1461 keV yields a coefficient needed to
convert all the relative efficiencies to absolute efficiencies The absolute efficiency curve in
Figure 39 was generated using this procedure for the Kloof sample In the same manner
absolute efficiency curves were generated for the other samples The parameterisation of
absolute efficiencies (ε = a Eb) is given in each relative efficiency plot
56
The graphs of ln(Energy) versus ln(Relative efficiency) for other samples follow from Figures
310 to 313
-12
-1
-08
-06
-04
-02
0
02
56 58 6 62 64 66 68 7
ln(E)
ln(R
elat
ive
effic
ienc
y)
232Th
a = 50708 b = -08572
Figure 36 Fit of relative efficiency with parameters a amp b generated (for a Kloof sample) asdetermined from lines associated with the decay of 232Th Values were normalized relative tothe 338 keV line
57
-15
-1
-05
0
05
1
0 2 4 6 8 10
ln(E)
ln(r
elat
ive
effic
iem
cy)
a = 4289 b = -07392
Figure 37 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Kloof sample number 6)
002040608
112141618
0 500 1000 1500 2000 2500
Energy (keV)
Rel
ativ
e ef
ficie
ncy
U(238)Th(232)K(40)
Figure 38 A relative efficiency curve for the sample number 6 (Kloof sample)
58
00005001
0015002
0025003
0035004
0045
0 500 1000 1500 2000 2500
Energy (keV)
Abs
olut
e ef
ficie
ncy
U(238)
Th(232)
K(40)
Figure 39 A typical absolute efficiencies versus energy graph for various selected lines associated with nuclides 238U 40K and 232Th for sample number 6b (Kloof sample)
59
-15
-1
-05
0
05
1
0
ln(r
elat
ive
effic
ienc
y)
U(238)Th(232)
Figure 310 A determined fromnumber 17A)
-25
-20
-15
-10
-50
5
10
15
20
25
0
ln(r
elat
ive
effic
ienc
y)
Figure 311 A
determined from
a = 45254 b = -07623
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (Westonaria sand sample
U (238)T(232)
a = 48294 b = -0772
1 2 3 4 5 6 7 8 9
ln(E)
fit of relative efficiency data with parameters a amp b generated as
lines associated with 238U and 232Th decay (West coast sand sample)
60
-2
-15
-1
-05
0
05
1
0 1 2 3 4 5 6 7 8 9
ln(E)
ln(r
elat
ive
effic
ienc
y)U(238)
Th(232)
a = 42021 b = -07252
Figure 312 A fit of relative efficiency data with parameters a amp b generated as determined from lines associated with 238U and 232Th decay (Brick clay)
- 2 5
- 2
- 1 5
- 1
0 5
0
0 5
1
1 5
0-
ln(r
elat
ive
effi
cien
cy) U ( 2 3 8 )
T h ( 2 3 2 )
Figure 313 Adetermined from
a = 51057 b = -08147
2 4 6 8 1 0
ln(E)
fit of relative efficiency data with parameters a amp b generated as lines associated with 238U and 232Th decay (thorium + stearic sample)
61
312 Treatment of uncertainties in WA
The uncertainties contributing to the results in this method were propagated throughout by
adding them all in quadrature combinations An example of the uncertainty due to background
subtraction is as follows
from equation 35 it follows that
22 )()( ibiNibiNiN CCCC ∆+∆plusmnminus=C prime
where 22 )()( ibiNiN CCC ∆+∆=prime∆ (311)
prime∆ iNC is an uncertainty in the background subtracted net counts and are
uncertainties due to counting statistics for net and background counts
iNC∆ ibC∆
Now to get to the relative efficiency the background subtracted net counts will be divided by
the branching ratio (which also has its specified uncertainty from the manual [EML97]) This
now yields the following
b
iNrel r
C prime=ε (312)
Therefore the uncertainty in 312 will then be given by
22
∆+
prime
prime∆=∆
b
b
iN
iNrelrel r
r
C
Cεε (313)
62
The uncertainties from the live time and the sample mass were almost negligible and therefore
were treated as constants
Furthermore from equation 36
baE=ε
the relative uncertainties include the uncertainty in parameters a and b as follows
22
22
2 bb
aa
∆
partpart
+∆
partpart
=∆εεε (314)
This reduces to
(315) ( )[ ] 2222 ln)( baEEaE bb ∆+∆=∆ εε
which is the uncertainty in the relative uncertainties (ignoring co-variances)
Although the uncertainties in a and b were determined these uncertainties were not
propagated in this study
The activity concentrations presented in chapter 4 by this method are calculated from
intensities of several γ-rays emitted by parent nucleus and therefore grouped together to
produce a weighted average activity per nuclide
These weighted average activity concentrations in the samples analysed (using the WA
method) are presented in Tables 41 to 49 in chapter 4
The equations in appendix A were used in the treatment of uncertainties for this method see
also the statistics of counting described in Appendix A2 to find out how weighted averages
are arrived at
63
32 Full Spectrum Analysis
The important aspects to consider in this method are the use of its full-spectral shape and the
so-called standard spectra in the calculation of the activity concentrations of natural
radionuclides 238U 232Th and 40K [Hen01] The total spectrum comprises a background
spectrum and the responses to the activity concentrations of the natural radionuclides These
responses are considered to be proportional to the activity concentrations and require detailed
knowledge of the standard spectra Each of the spectra corresponds to the response of the
detector in a particular geometry for a sample containing 1Bqkg of each nuclide [Hen03]
All the spectral features including the inclusion of the Compton continuum in the spectrum
makes this method a good one in the data analysis and this contributes to the reduction of the
statistical uncertainty in the data especially for relatively short measurements since in principle
more time is required for the accumulation of data with small uncertainties
321 Derived standard spectra
Energy calibration was done according to the calibration process already discussed in section
214 These energy calibrations were done independently for each standard spectrum
In general the relation between energy and channel number of the sample spectra measured
deviates from that of the standard spectra These deviations can be attributed for example to
temperature variations which consequently result in the varying in time of the relation
between energy and channel number [Kno79]
64
To take the differences in amplifications for samples taken at different times into account all
spectra were re-binned5 using the energy calibration parameters so that each spectrum has the
same channelenergy relationship chosen as 1 keV per channel for convenience
M(j)
j S S
Figure 314 The contents of channels of the measured spectrum S redistributed to the re-binned spectrum S
The channel i of the re-binned spectrum S can be mapped onto the channels j of the
measured spectrum S with a function i = M(j) and the contents of the channels of the
measured spectrums can then be redistributed over the channels of the re-binned spectrum S
[Sta97] A small FORTRAN program was developed to execute such a task see (appendix B)
The re-binning exercise is needed to try and correct for the small differences in the energy
calibration between the standard and sample spectra
The spectra were normalized by dividing each channel by the product of source mass live time
and activity concentration The flow chart in Figure 315 shows a summary of how standard
spectra were obtained
65
5 channels converted to equivalent energy values per bin
For a particular spectrum divide eachby a product of source mass livetime and activity concentration
Standard spectrum
Re-bin each spectrum such that 1 channel = 1 keV
Energy calibrate each spectrum
Measure reference spectrumsource on HPGe
Figure 315 Flow chart showing how a standard spectrum was obtained
Figures 316 317 and 318 show the re-binned spectra for the three standard sources used
which are those for the 238U 232Th and40K respectively
66
00001
001
1
100
50 100 150 200 250 300 350 400 450 500
00001
001
100
500 550 600 650 700 750 800 850 900 950 1000
00001000100101
1
100
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
00001000100101
1
100
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
00001000100101
1
100
2000 2100 2200 2300 2400 2500 2600
1
67
10
10
Cou
nts
seco
nd (
log
scal
e)
10
Figure 316 The background subtracted re-binned standard spectrum for uranium
Energy (keV)
100E-03100E-02100E-01100E+00100E+01100E+02
50 100 150 200 250 300 350 400 450 500
100E-03100E-02100E-01100E+00100E+01100E+02
500 550 600 650 700 750 800 850 900 950 1000
100E-03
100E-02
100E-01
100E+00
100E+01
100E+02
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450 1500
100E-03100E-02100E-01100E+00100E+01100E+02
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
100E-03100E-02100E-01100E+00100E+01100E+02
2000 2100 2200 2300 2400 2500 2600
68
Cou
nts
seco
nd (
log
scal
e)
Figure 317 The background subtracted re-binned standard spectrum for thorium
Energy (keV)
69
001
01
1
10
50 550 1050 1550 2050 2550
0001
4
Cou
nts
seco
nd (
log
scal
e)
1 32
Energy (keV)
Figure 318 The background subtracted re-binned standard spectrum for potassium (Peaks labeled 1 2 3 and 4 are the full-energy peaks 438 keV due to second escape peak the annihilation peak (511 keV) the first escape peak (950 keV) and the 1461 keV due to 40K respectively
322 Chi-square minimization technique
Once the fixed relation between energy and channel number has been obtained the least-
square method is used to find the optimal activity concentrations using the Physica software
[Chu94] In this method the activity concentrations will follow from a fit of the calculated
standard spectrum to the measured one [Hen01] In the FSA the measured spectrum Y is
described as the sum of the standard spectra Xj multiplied by the activity concentrations Cj for
the individual radionuclides The background was subtracted from the individual spectrum in
each measurement before fitting
If a complex spectrum has to be decomposed into j known components the intensity Cj of
each component relative to that of its standard is determined by the requirement that
sum sum=
=
minus
minus=
hecn
leci jjj iXCiYiw
mnred
22 )()()(1χ (316)
should be a minimum [Qui72Pre92Hen10] Y (i) and Xj(i) are counts in channel i recorded
under the same experimental conditions j represents the number of standard spectra used
with jmax= 3 since three standards sources (238U 232Th and 40K ) were used in this study i =
lec and n = hec are low energy and high energy cut-offs used during the minimization
procedure The factor n-m represents the number of degrees of freedom where n is the
number of data points and m the number of free parameters The choice of n-m assumes all
the channels to be independent [Hen03] The weight factor is given as the inverse of the
square of the standard deviation The standard deviations for each source were added in
quadrature combinations to the samples standard deviation
)(iw
The important part of equation 316 for FSA is in the definition of the weights w(i) and this
will be discussed next
70
Definition of weights
Let
AAA ii ∆=plusmnand
BB ii ∆=plusmnΒ
represent gross counts in the ith channel of the sample and background spectrum respectively
Now the real count rate obtained after background subtraction is given by
22
∆+
∆plusmn
minus=
BAB
i
A
i
tB
tA
tB
tA
CR (317)
Were tA and tB are live times for the measured sample and background respectively The
uncertainty in the background-subtracted count rates is given by
22
∆+
∆=
B
i
A
ii t
BtA
σ (318)
The errors in (316) were added in quadrature combinations to give the total standard
deviation
( )2222
2
22
2
22
2
2
2
222 1 ThUK
B
i
U
UU
Th
ThTh
S
S
K
KKi CCC
tB
tA
CtA
CtA
tAC +++
∆+
∆+
∆+
∆+
∆=σ (319)
71
where the subscripts S K Th and U are the sample potassium thorium and uranium spectra
respectively and with t being the live time associated with the spectra The background B was
subtracted from each spectrum that is in all three standard spectra plus the sample one
The weighting is then given by the inverse of the square of the standard deviations as follows
w(i) = 1σ i2 (320)
Therefore [ ]sum
=
+= 3
1
22 )()]([
1)(
jXjS iCi
iw
jσσ
(321)
where Sσ is the standard deviation for the sample measurement
The minimization is such that
02
=partpart
jcχ ( for all j ) (322)
72
Several measurements on different samples were made with both methods and the Tables 23
and 26 show the information regarding such samples
The fit is reasonably good in general except for low energy cut-off energies less than about
300 keV This is due to the self-absorptions at lower energies as discussed later in chapter 4
The χ2reduced has been calculated for low energy cut-off energies ranging from 50 keV up to
300 keV (in steps of 50 keV) and with a high energy cut-off of 2650 keV These χ2reduced values
obtained are shown in Figure 319 for the different cut-off energies
0
5
10
15
20
25
0 100 200 300 400 500 600 700
Energy(keV)
redu
ced
chi-s
quar
e
Figure 319 Reduced chi-square (measure of the goodness of fit) for FSA fit of Westonaria sample as a function of lower energy cut-off (lec) (see equation 316)
73
001
01
1
10
50 100 150 200 250 300
000001
00001
0001
001
01
50 100 150 200 250 300
Measured Fit
Cou
nts
seco
nd
(b)
(a)
Figure 320 The background subtracta long measurement (below 300 keV)
Energy (keV)
ed Kloof (a )and West Coast(b) sample spectra and their fit for
74
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
50 100 150 200 250 300
(d)
0001
001
01
1
10
50 100 150 200 250 300
(d)
(c)
Energy (keV)
Figure 321 The background subtracted Brick clay (c) and thorium sample(d) spectra andtheir fit for a long measurement (below 300 keV )
75
0001
001
01
1
10
300 350 400 450 500
0001
001
01
1
10
500 600 700 800 900 1000
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 322 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 300 and 1000 keV)
76
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
00001
0001
001
01
1
10
1500 1600 1700 1800 1900 2000
Energy (keV)
Measured Fit
Cou
nts
seco
nd
Figure 323 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 1000 and 2000 keV)
77
00000100001000100101
110
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Cou
nts
seco
nd
Energy (keV)
Figure 324 The background subtracted Kloof sample spectrum for a long measurement and the fit (forenergies between 2000 and 2650 keV)
78
001
01
1
50 100 150 200 250 300
C
ount
sse
cond
s
Measured Fit
Energy (keV)
001
01
1
10
300 350 400 450 500
Figure 325 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 50 and 500 keV)
79
Energy (keV)
Cou
nts
seco
nd
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
Energy (keV)
0001
001
01
1
10
1000 1100 1200 1300 1400 1500
Figure 326 The background subtracted Kloof sample spectrum for a short measurement and the fit(for energies between 500 and 1500 keV)
80
00000100001000100101
110
2000 2200 2400 2600
Energy (keV)
Cou
nts
seco
nd (
log
scal
e)
00001000100101
110
1500 1600 1700 1800 1900 2000
Figure 327 The background subtracted Kloof sample spectrum for a short measurement and the fit (forenergies between 2000 and 2650 keV)
81
Measured Fit
0001
001
01
1
10
500 600 700 800 900 1000
0001
001
01
1
10
300 350 400 450 500
Energy (keV)
Cou
nts
seco
nd
Figure 328 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 300 and 1000 keV)
82
Measured Fit
0001
001
01
1
1500 1600 1700 1800 1900 2000
0001
001
01
1
1000 1100 1200 1300 1400 1500
Energy (keV)
Cou
nts
seco
nd
Figure 329 The background subtracted thorium + stearic acid sample spectrum and the fit for along measurement (for energies between 1000 and 2000 keV)
83
00001
0001
001
01
1
2000 2100 2200 2300 2400 2500 2600
Measured Fit
Energy (keV)
Cou
nts
seco
nd
Figure 330 The background subtracted thorium + stearic acid sample spectrum and the fitfor a long measurement (for energies between 2000 and 2600 keV)
The fit is not good for the energies ranging from 50 keV to about 300 keV as is witnessed
from Figures 320 b and 321c This is due to self-absorption effects a phenomenon to be
discussed in the next chapter
In order to see how good a fit is two energy lines from two samples of different densities were
chosen The Figure 331 shows how good the fit is at the 609 keV (214Bi line) for the Kloof
sample while Figure 332 shows the fit for the 583 keV (208Tl line) from the thorium + stearic
acid sample
84
0
05
1
15
605 606 607 608 609 610 611 612
Energy ( KeV)
Cou
nts
seco
nd FitMeasured
Figure 331 A typical 609 keV peak due to 214Bi (for Kloof sample number 6) with a fit
0
05
1
15
580 581 582 583 584 585
Energy(keV)
coun
tss
econ
ds FitMeasured
Figure 332 A typical 583 keV peak due to 208Tl (for thorium + stearic acid sample
number 5) with a fit
85
323 Treatment of uncertainties for FSA
The uncertainties in Cj from equation 316 were obtained using the Gauss-Newton method
This method proceeds by iterative means to calculate ∆C such that
CCC ∆+= 01 (323)
the aim of this method is to find a ∆C such that C1 is a better approximation to the best least
square fit solution [Chu94Pre92] If the method is repeated iteratively the series C1C2 C3
will converge until the sum of the squares of the differences between the data points and the
function being fitted is a minimum
To accomplish this the function f (xC) is expanded in a Taylors series about the point C = C0
and only the first order terms are kept [Chu94]
C
CCxfCxfCxf
part∆part
+=)(
)()( 00 (324)
The equation above is an exact equality when f is linear in the parameters C that is all second-
order and higher derivatives are zero
The problem then is to minimize
2
00
11
2 )()(
part
∆partminusminus= sumsum
== CCCxf
Cxfyw kkk
N
kk
N
kkδ
for a system with M parameters this becomes
2
1
00
1
)()(
part
∆partminusminus= sumsum
==
M
j j
jkkk
N
kk C
CCxfCxfyw (325)
86
If the above expression is differentiated with respect to each of the unknowns ∆Cj and the
resulting expressions are set to zero the problem reduces to solving the matrix equation
rCA =∆ where A = (aij) r = (ri)
and j
kN
K i
kkij p
Cxfp
Cxfwa
partpart
partpart
= sum=
)()( 0
1
0 for i j = 1M (326)
[ ]i
kkk
N
kki c
CxfCxfywr
partpart
minus= sum=
)()( 0
01
for i = 1M (327)
sum=
N
kk
1
2δ is zero at the minimum provided the Gauss-Newton method is locally convergent
The advantage of this method is that linear least squares problems are solved in one iteration
An indication of the accuracy of the fit is displayed in the output of the Physica program as E1
and E2 where
iib=1Ε for i=1M (328)
where bii are the diagonal elements of B = A-1 the inverse of the matrix A B is called the
covariance matrix and E1 is referred to as the root mean square statistical error of estimate
The root mean square total error of estimate or standard error is displayed under E2 where
Ε for j = 1M (329) 21
1
212 )(
minusΕ= sum
=
N
KkK MNw δ
Therefore the accuracy of the parameters in a linear fit is
Ci plusmn E2i for j = 1M
87
Chapter 4
Results and Discussion
In this chapter the activity concentrations of 238U 232Th and 40K in the samples analysed as
derived from Windows Analysis and Full Spectrum Analysis are presented and compared
41 WA results
The activity concentrations and associated uncertainties as derived using long live time
measurements are given in Tables 41 - 45
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 3054 plusmn 50 00410214Pb 2951 3289 plusmn 29 00295214Pb 3520 3337 plusmn 27 00260214Bi 9340 2768 plusmn 102 00129214Bi 11203 2957 plusmn 35 00113214Bi 12381 2991 plusmn 65 00105214Bi 13777 3289 plusmn 83 000980214Bi 17655 3221 plusmn 38 000821214Bi 22041 3192 plusmn 63 000700
320 plusmn 5
232Th series 228Ac 3384 251 plusmn 15 00267212Bi 7273 193 plusmn 30 00154228Ac 7948 195 plusmn 51 00145208Tl 8603 264 plusmn 63 00137228Ac 9112 187 plusmn 09 00131228Ac 9668 228 plusmn 04 00126
21 plusmn 1
40K Series 40K 14608 244 plusmn 4 000940
Table 41 The activity concentration results for the Kloof sample number 6 (longmeasurement) by WA method
88
Nuclide Energies (keV)
Activity Concentration
(Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration
(Bqkg) 238U series 226Ra238U 1861 2821 plusmn 47 00451214Pb 2951 2520 plusmn 12 00318214Pb 3520 2691 plusmn 16 00278214Bi 9340 2359 plusmn 78 00132214Bi 11203 2519 plusmn 27 00115214Bi 12381 2552 plusmn 58 00107214Bi 13777 2992 plusmn 64 000983214Bi 17655 2752 plusmn 25 000815214Bi 22041 2822 plusmn 49 000687
263 plusmn 4
232Th series 228Ac 3384 191 plusmn 09 00286212Bi 7273 240 plusmn 20 00160228Ac 7948 236 plusmn 27 00149208Tl 8603 165 plusmn 34 00141228Ac 9112 153 plusmn 09 00135228Ac 9668 215 plusmn 12 00129
19 plusmn 1
40K Series 40K 14608 230 plusmn 3 000940
Table 42 The activity concentration results for the Westonaria sample number 17A (long measurement) by WA method
89
Nuclide Energies (keV)
Activity Concentr
ation (Bqkg)
Full-energy peak
Absolute efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 64 plusmn 13 00558214Pb 2951 40 plusmn 05 00375214Pb 3520 53 plusmn 03 00321214Bi 9340 63 plusmn 10 00138214Bi 11203 47 plusmn 27 00118214Bi 12381 43 plusmn 19 00108214Bi 13777 45 plusmn 32 000989214Bi 17655 51 plusmn 10 000799214Bi 22041 42 plusmn 21 000659
53 plusmn 03
232Th series 228Ac 3384 28 plusmn 06 00333212Bi 7273 51 plusmn 15 00172228Ac 7948 92 plusmn 17 00159208Tl 8603 102 plusmn 24 00148228Ac 9112 43 plusmn 04 00141228Ac 9668 37 plusmn 09 00134
36 plusmn 03
40K Series 40K 14608 593 plusmn 15 000940
Table 43 The activity concentration results for (long measurement) the West Coast sample number 3 by WA method
90
Nuclide Energies (keV)
Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 314 plusmn 29 0064 214Pb 2951 297 plusmn 20 00417 214Pb 3520 313 plusmn 21 00353 214Bi 9340 221 plusmn 65 00143 214Bi 11203 369 plusmn 32 00120 214Bi 12381 273 plusmn 49 00110 214Bi 13777 365 plusmn 58 000993 214Bi 17655 405 plusmn 37 000789 214Bi 22041 450 plusmn 64 000641
30 plusmn 14
232Th series 228Ac 3384 450 plusmn 30 00367 212Bi 7273 658 plusmn 53 00180 228Ac 7948 622 plusmn 58 00166 208Tl 8603 599 plusmn 64 00154 228Ac 9112 575 plusmn 43 00146 228Ac 9668 614 plusmn 47 00138
55 plusmn 4
40K Series 40K 14608 216 plusmn 3 000940
Table 44 The activity concentration results for (long measurement) the brick clay sample number 6 by WA method
91
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 304 plusmn 79 00382 214Pb 2951 134 plusmn 23 00279 214Pb 3520 137 plusmn 17 00248 214Bi 9340 194 plusmn 135 00128 214Bi 11203 129 plusmn 27 00112 214Bi 12381 -09 plusmn -78 00105 214Bi 13777 -09 plusmn -117 000978 214Bi 17655 73 plusmn 149 000827 214Bi 22041 129 plusmn 27 000710
13 plusmn 1
232Th series 228Ac 3384 4566 plusmn 48 00255 212Bi 7273 5338 plusmn 88 00151 228Ac 7948 4347 plusmn 121 00142 208Tl 8603 5071 plusmn 125 00135 228Ac 9112 4490 plusmn 36 00129 228Ac 9668 4414 plusmn 46 00125
469 plusmn 8
40K Series 40K 14608 31plusmn5 000940
Table 45 The activity concentration results for (long measurement) the thorium + stearic acid sample number 5 by WA method
92
The activity concentrations and associated uncertainties as derived using short live time
measurements are given in Table 46 - 49
Nuclide Energies
(keV) Activity Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 2508 plusmn 133 00443214Pb 2951 2655 plusmn 52 00317214Pb 3520 2883 plusmn 40 00278214Bi 9340 2710 plusmn 278 00137214Bi 11203 2413 plusmn 87 00120214Bi 12381 2351 plusmn 186 00111214Bi 13777 2934 plusmn 235 00103214Bi 17655 2837 plusmn 43 000859214Bi 22041 2854 plusmn 165 000730
277 plusmn 5
232Th series 228Ac 3384 207 plusmn 42 00369212Bi 7273 253 plusmn 88 00287228Ac 7948 79 plusmn 159 00164208Tl 8603 162 plusmn 184 00154228Ac 9112 188 plusmn 26 00145228Ac 9668 264 plusmn 49 00139
21 plusmn 2
40K Series 40K 14608 244 plusmn 11 000986
Table 46 The activity concentration results (short measurement) for the Kloof sample number6 by WA method
93
Nuclide Energies (keV)
Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity
Concentrations (Bqkg)
238U series 226Ra238U 1861 263 plusmn 13 00451214Pb 2951 247 plusmn 5 00318214Pb 3520 270 plusmn 4 00278214Bi 9340 260 plusmn 26 00132214Bi 11203 222 plusmn 8 00115214Bi 12381 232 plusmn 16 00107214Bi 13777 204 plusmn 24 000983214Bi 17655 268 plusmn 7 000815214Bi 22041 283 plusmn 15 000687
257 plusmn 6
232Th series 228Ac 3384 48 plusmn 42 00286212Bi 7273 184 plusmn 78 00160228Ac 7948 48 plusmn 150 00149208Tl 8603 151 plusmn 3 00141228Ac 9112 213 plusmn 5 00135228Ac 9668 50 plusmn 14 00129
14 plusmn 2
40K Series 40K 14608 216 plusmn 11 000940
Table 47 The activity concentration results for the Westonaria sample number 17A (short measurement) by WA method
94
Nuclide Energies (keV)
Activitiy Concentration
(Bqkg)
Full-energy peak Absolute
efficiency
Weighted Average Activity
Concentration (Bqkg)
238U series 226Ra238U 1861 80 plusmn 27 00450214Pb 2951 42 plusmn 10 00317214Pb 3520 47 plusmn 07 00277214Bi 9340 75 plusmn 60 00132214Bi 11203 56 plusmn 14 00115214Bi 12381 -04 plusmn -62 00107214Bi 13777 -04 plusmn -69 000982214Bi 17655 67 plusmn 11 000814214Bi 22041 11 plusmn 51 000688
52 plusmn 05
232Th series 228Ac 3384 42 plusmn 10 00286212Bi 7273 44 plusmn 20 00160228Ac 7948 15 plusmn 48 00149208Tl 8603 49 plusmn 48 00140228Ac 9112 46 plusmn 08 00134228Ac 9668 56 plusmn 13 00129
46 plusmn 05
40K Series 40K 14608 54 plusmn 4 000940 Table 48 The activity concentration results for the West Coast sample number 3 (short measurement) by WA method
95
Nuclide Energies
(keV) Activitiy Concentration (Bqkg)
Full-energy peak Absolute efficiency
Weighted Average Activity Concentration (Bqkg)
238U series 226Ra238U 1861 329 plusmn 65 00532214Pb 2951 320 plusmn 23 00365214Pb 3520 340 plusmn 17 00318214Bi 9340 288 plusmn 159 00142214Bi 11203 214 plusmn 30 00123214Bi 12381 423 plusmn 97 00113214Bi 13777 590 plusmn 35 00103214Bi 17655 378 plusmn 40 000845214Bi 22041 199 plusmn 144 000704
32 plusmn 2
232Th series 228Ac 3384 416 plusmn 32 00326212Bi 7273 618 plusmn 70 00174228Ac 7948 318 plusmn 58 00162208Tl 8603 469 plusmn 118 00152228Ac 9112 583 plusmn 14 00145228Ac 9668 585 plusmn 30 00138
55 plusmn 3
40K Series 40K 14608 220 plusmn 9 000986
Table 49 The activity concentration results for the brick clay sample number 6 (short measurement) by WA method
96
42 FSA results
The FSA results were obtained using a low-energy cut-off of 300 keV for long measurements
and 400 keV for short measurements (see also Figure 319) The fits on which results are based
are shown in Figures 320 to 332 The derived activity concentrations from long and short
measurement are given in Tables 410 to 411 respectively
Sample description
Activity Concentration in (Bqkg) (for long measurement) with a cut off energy of 300 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
222 plusmn 5 288 plusmn 04 542 plusmn 05 19
5 thorium+stearic acid
1 plusmn 4 08 plusmn 05 3954 plusmn 03 196
Table 410 The FSA results (long measurement) uncertainties and the associated chi-square per degree of freedom obtained using cut-off energy of 300 keV for the samples indicated
97
Sample description
Activity Concentration in (Bqkg) (for short measurements) with cut off energy of 400 keV Full Spectrum Analysis (FSA)
S
Type of Sample
40K 238U 232Th
χ2ν
D3 West Coast sand
356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria sand
250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 411 The FSA results (short measurements) uncertainties and the associated chi-
square per degree of freedom obtained using a cut off energy of 400 keV for the samples
indicated
98
43 Comparison of WA and FSA results
The weighted average WA results and FSA results are tabulated and juxtaposed in Tables 413
and 414 for long and short measurements respectively A comparison of the results is shown
graphically in Figures 41 to 412
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for long measurements with cut-off energy of 300 keV
Windows Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast sand
593 plusmn 15 53 plusmn 03 36 plusmn 03 61 plusmn 3 40 plusmn 01 22 plusmn 03 16
17A Westonaria sand
230 plusmn 3 263 plusmn 4 19 plusmn 1 227 plusmn 4 232 plusmn 1 18 plusmn 02 35
6B Kloof mine sand
244 plusmn 4 320 plusmn 5 21plusmn 1 242 plusmn 4 272 plusmn 1 194 plusmn 02 23
D6 Brick clay
216 plusmn 3 30 plusmn 14 55 plusmn 4 222 plusmn 5 288 plusmn 04 542 plusmn 05 19
Table 412 The activity concentrations from the two methods of analysis for long measurement
99
iThemba LABS ERL Soil Measurements
Activity Concentration in (Bqkg) for short measurements at the cut-off energy 400 keV
Window Analysis (WA)
Full Spectrum Analysis (FSA)
Ref no
Sample Type
40K 238U 232Th 40K 238U 232Th
χ2
red
D3 West Coast Sand
54 plusmn 4 52 plusmn 05 46 plusmn 05 356plusmn 33 07plusmn 03 33plusmn 03 19
17A Westonaria Sand
216 plusmn 11 257 plusmn 6 14 plusmn 2 250 plusmn 10 241 plusmn 2 21plusmn 1 22
6B Kloof mine Sand
244 plusmn 11 277 plusmn 5 21 plusmn 2 240 plusmn 9 237plusmn 2 20plusmn 1 18
D6 Brick clay
220 plusmn 9 32 plusmn 2 55 plusmn 3 220 plusmn 7 31 plusmn 1 59plusmn 1 14
Table 413 The activity concentrations from the two methods of analysis for short measurements
The long measurement results of these two methods (WA and FSA) were plotted on the
histograms to show the variations in activity concentrations for various samples The percent
uncertainties were also shown see Figures 41 to 412 The notations WAL WAS FSAL and
FSAS are defined as follows
WAL = Window Analysis Long (for long measurement)
WAS = Window Analysis Short (for short measurement)
FSAS = Full Spectrum Analysis (for short measurement)
FSAL = Full Spectrum Analysis (for long measurement)
100
0
50
100
150
200
250
300
K U Th
nuclide
Act
ivity
(Bq
kg)
WAL
FSAL
Figure 41 The comparison of the three nuclides for Westonaria sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 42 The comparison of the three nuclides for Westonaria sand (short measurement) in both window and FSA analyses
e s t o n a r i a s a m p l e
02468
1 01 21 41 6
0 1 2 3 4n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
W
40K 232Th 238U
Figure 43 The percentage uncertainties for activity concentrations as a function of nuclide for Westonaria sand sample
101
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
3 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A LF S A L
Figure 44 The comparison of the three nuclides for Kloof sand (long measurement) in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
3 0 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 45 The comparison of the three nuclides for Kloof sand (short measurement) in both window and FSA analyses
K l o o f
0
2
4
6
8
1 0
1 2
0N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 46 The percentage uncertainties for activity concentrations as a function of nuclide for Kloof sand sample
102
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l id e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 47 The comparison of the three nuclides for west coast (long measurement) in both window and FSA analyses
0
1 0
2 0
3 0
4 0
5 0
6 0
7 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 48 The comparison of the three nuclides for West coast (short measurement) in both window and FSA analyses
W e s t C o a s t
05
1 01 52 02 53 03 54 04 5
0
N u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 49 The percentage uncertainties for activity concentrations as a function of nuclide for West Coast sand sample
103
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)W A LF S A L
Figure 410 The comparison of the three nuclides for brick clay for long measurement in both window and FSA analyses
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
K U T
N u c l i d e
Act
ivity
Con
cent
ratio
n (B
qkg
)
W A SF S A S
Figure 411 The comparison of the three nuclides for brick clay for short measurement in both window and FSA analyses
B r ic k c l a y
0
1
2
3
4
5
6
7
0
n u c l i d e
un
cert
aint
y
W A LW A SF S A LF S A S
41 2 3 40K 232Th 238U
Figure 412 The percentage uncertainties for activity concentrations as a function of nuclide for brick clay sample
104
0
50
100
150
200
250
300
350
0 50 100 150 200 250 300 350
Activity concentration via FSA in(Bqkg)
Act
ivity
con
cent
ratio
n vi
a W
AM
in(B
qkg
)
KUTh
R2 = 0932
Figure 413 A graph showing correlation of activity concentration determined using the WAM vs FSA on average the two methods agree well within the correlation coefficient of 0932 as depicted on Figure
105
The deviation between results from the two methods for 238U concentrations (long
measurements) was found to be about 18 for the Kloof 13 for Westonaria 4 for brick
clay and 25 for West coast sand (see Figure 414) The deviations between results from long
measurement for 232Th yielded 6 for Kloof sample 5 for Westonaria sample 1 Brick
clay and 63 for West coast sand (see Appendix E for the ratios presented) The deviations
are consistently higher by these percentages for WA than FSA This trend has to do with the
fact that the uranium and thorium sources (used for FSA method) did not fill 1 litre Marinelli
beaker This could be attributed to the self-absorption effects which vary as a function of
volume The evidence of this is presented later in section 45 An attempt to study the volume
and density effects was made through the use of the Sima model [Sim92] and the results from
this modelling are presented in section 45 (see also appendix D)
The low activity West coast sample resulted in very high uncertainty in 238U activity
concentrations (see Figure 49 and 415) This was especially true for the FSA method which
has proved to find it difficult to determine the activity concentrations of low activity nuclides
This is because of the contribution of the background counts which at times are higher than
net counts especially in short measurements thus yielding a poor fit
1
0 8
0 9
1
1 1
1 2
1 3
1 4
0
s a
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
2 1
S
ratio
of a
ctiv
ity
conc
entr
atio
ns
232Th
238U
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
s a
ratio
s of
act
ivity
conc
entr
atio
ns
40K
Figure 414 The activity concentration ratipotassium long measurement for various sample
5
1 2 3 4 Kloof Westonaria Brick clay West Coast
m p le s
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
a m p le
5
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
06
m p le s
os (WAFSA) of uranium thorium ands
0 80 8 5
0 90 9 5
11 0 5
1 11 1 5
1 2
0
S a m p le
conc
entr
atio
ns
0 50 70 91 11 31 51 71 92 1
5
s a m p le s
ratio
of a
ctiv
ity
conc
entr
atio
ns
0 7
0 9
1 1
1 3
1 5
1 7
1 9
5
s a m p le
ratio
of a
ctiv
ity
conc
entr
atio
ns238U
ratio
of a
ctiv
ity
1 2 3 Kloof Westonaria Brick clay 4
232Th
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
40K
0 1 2 3 4 Kloof Westonaria Brick clay West Coast
Figure 415 The activity concentration ratios (WAFSA) of uranium thorium and potassiumfrom short measurements for various samples The ratio for the 238U (West coast sample) isnot shown
107
Method Advantages Disadvantages Significant source of
uncertainty
WA
No correction for
self-absorption
effects needed and
one standard source
required (ie KCl)
Some lines
prone to
summing
effects
Short time
measurements
FSA
Short
Measurements and
No worry about
Summing effects
Three standard
sources required
Some resolution may
be lost during the
rebinning of the
spectrum
Table 414 Advantages and disadvantages for the two methods including the best measurement times required for the activity concentration determination For samples where 40K and 232Th have the same activity concentration the 40K line is ten times
stronger than the 232Th line [Dem97] In case the 232Th content is several orders of magnitude
larger it becomes virtually impossible to determine the 40K content accurately since the peaks
are very close to each other
The results of the high thorium content are tabulated in Table 415 for the evidence of that
For the FSA this summing effect is not a problem since all the peaks are considered for the
determination of the content of these two nuclides that is the 40K and 232Th In the WA the
14608 keV peak is confused with the 14592 keV one
108
40K 1 plusmn 4 31 plusmn 5
238U 08 plusmn 05 12 plusmn 1
3954 plusmn 03 469 plusmn 8
FSA activity concentrations (Bqkg)
WA activity concentrations (Bqkg) Nuclide
232Th
Table 415 The results of sample 5 a high thorium content sample
050
100150200250300350400450
K U Th
Nuclide
Act
ivity
con
cent
ratio
ns
(Bq
kg)
WAFSA
Figure 416 Activity concentration of nuclides for the high thorium content sample
109
bull Thorium sample
As discussed in chapter 2 the thorium + stearic sample was made up using stearic acid and
IAEA reference material (RGTh-1) having an activity concentration of 3250 Bqkg (see Table
24) The sample was prepared in such a way that the activity concentration of 232Th in the
sample was 5000 plusmn 05 Bqkg [Jos04]
The WA and FSA results for this sample allow us to compare WA and FSA derived 232Th
concentrations with what is expected As can be seen in Table 415 the WA yields a result of
469 Bqkg which is 9 smaller than expected The FSA gives the result of 395 Bqkg which
is 21 smaller than expected It is clear from Figure 331 that the counts in FSA fit spectrum
are generally higher than in the measured spectrum in regions outside photo peaks This
happens since the full-energy peak efficiency (also termed the photopeak efficiency) is higher
in the case of thorium sample since it has such a low density (~07 gcm3) resulting in a higher
peak to total ratio (ie relatively more counts inside than outside the photopeak) Since the
FSA procedure involves the minimisation of reduced chi-square the photopeaks are fitted
preferentially since a misfit in the region of a photopeak contributes significantly to reduced
chi-square
44 Discussion of WA Results
Counting uncertainties in environmental measurements are usually high such that coincident
summing corrections might be neglected [Gar01] But the lines that are prone to coincident-
summing6 such as the 609 keV were still avoided for purpose of reducing the systematic error
At present the ERL at iThemba LABS has a study going on about the prominent lines that are
prone to the coincidence-summing problem Other lines that are prone to the coincident-
6 This is when γ-rays are emitted in cascades from a nucleus and thus detected as one peak
110
summing like the 9112 keV 9690 keV from 228Ac and 7273 keV due to 212Bi might in future
be omitted [Gar01]
Accumulation of good statistics is required for the reliability of this method This was
demonstrated by comparing the uncertainties associated with measurements of varying
duration Results from shorter measurements were more uncertain than those from long ones
(see Figures 434649 and 412)
45 Discussion of FSA Results
Contrary to WA which only uses the data in a number of peaks for analysis FSA includes all
the spectral information in the data analysis The increased amount of information will
decrease the statistical uncertainties in the method thus giving this method an advantage
A practical problem of the FSA method might be the fact that small gain drifts may occur
resulting in the slight shifts and broadening of peaks
The results for this method are given for the cut-off energy of 300 keV for the long
measurement and for shorter measurement a lower cut-off energy of 400 keV and higher cut-
off energy of 1600 keV were used to exclude high energy background counts
This particular cut-off energy was chosen because of the poor fit as observed for low energies
(see Figures 320 and 321) This is reflected by the high values of χ2ν Figure 320 shows an
under-estimation of the measured spectrum with a chi-square 25 at the cut-off energy of 300
keV for a typical spectrum of Kloof sample 6 This under prediction at lower energies is
attributed to the self-absorption effects
From Table 24 it is clear that the Marinelli beakers fillings were not the same in volume
therefore this has a consequence on the measurement result due to these volume effects
which are related to self-absorption effects Figures 417 418 are the results showing the
transmission factors from the Sima calculations for various samples with different densities as
111
discussed in Appendix D while Figure 419 illustrates the volume effects for the uranium
source Debertin and Ren did a similar study [Deb89]
The poor reproduction of the FSA at energies less than 300 keV led to the studying of self-
absorption effects based on the Sima model
Self-absorption is the removal of γ-ray by material in the sample itself The effect increases for
lower energies (lt 300 keV) and with the atomic number of an element [Dem97]
Figures 417 and 418 show the results of the effect of density on the transmission factors (see
Appendix D for discussion on this) while Figure 419 shows the volume effect on the
transmission factors
112
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)Water(10)
Figure 417 Calculated transmission factors for different densities for a 1000 cm3
Marinelli as a function of γ-ray energy the legends shows the three reference sources
which are Potassium Chloride (KCl) uranium and thorium together with water and sand
at various densities (densities are shown by the numbers in brackets in the legends)
113
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500Energy in (keV)
Tran
smis
sion
Fac
tor
KCl(13)Sand(16)U(12)Th(13)H20(10)
Figure 418 The transmission factor for a 500 cm3 Marinelli at different densities as a function of γ-ray energy
114
0300
0400
0500
0600
0700
0800
0900
1000
0 500 1000 1500 2000 2500
Energy(keV)
Tran
smis
sion
Fac
tor
1000ml
500ml
Figure 419 The volume effect on the transmission factor for different fillings (with sand of density 16 gcm3) of Marinelli as a function of γ-ray energy
115
During the determination of the detection efficiency an error may occur due to the difference
in the densities of the standard source and the sample This effect can be verified from the
Figure 417 In this Figure it is clear that at lower energies samples with high densities such as
that of sand at 16 gcm3 get attenuated even more while the less dense samples such as water
at a density of 1 gcm3 get attenuated even less This trend is strong at energies less than 300
keV and at higher energies is almost negligible The other difference is due to the atomic
number this difference can be witnessed by the KCl which gets attenuated more at lower
energies than sand even if its density is less than that of sand This is simply because KCl has
an effective atomic number Z greater than that of SiO2 which is a major constituent of sand
Most of the photon attenuation occurs at low energy with Marinelli beaker filled to its full
capacity while at lower volumes there is less attenuation this is because it takes photons far
away from the detector even more time to arrive at the detection point and hence they get
attenuated on their path see Figure D1 (Appendix D) for the variations in the filling of the
beaker
The under prediction of the fit at the continuum for the FSA method of analysis is attributed
to this concept especially because the source volumes were plusmn 400ml for thorium and
uranium while samples had volumes close to 100 ml see Tables 23 and 24 for details
Different filling heights of the Marinelli beaker yield different effective thickness which were
used to calculate the transmission factors as in the Figure D1 (see also Table D1 in Appendix
D) The percentage difference according to these volume differences was calculated for full
(1000 ml) and half-full (500 ml) Marinelli beaker These percentage differences are plotted in
Figure 420
116
012345678
0 500 1000 1500 2000 2500
Energy (keV)
d
iffer
ence
Figure 420 The differences in the transmission factors of (Westonaria sand) with regard to full and half full Marinelli beaker as a fuction of energy
On interpolation the percent difference due to volume for the 14608 keV line was found to be
about 15 The percent difference D was calculated as follows
100 timesminus
=full
halffull
TTT
D (41)
where Tfull and Thalf are transmission factors for a full and half-full Marinelli beakers
respectively These self-absorption effects are of major concern this can be seen in Figure 319
for energies below 300 keV therefore another approach of defining these effects will be to
describe the probabilities of the interaction of γ-ray with matter inside both the detector and
Marinelli beaker This will be done by following a γ-ray photon of a particular energy inside the
Marinelli beaker until the detector absorbs it Two possibilities were taken into consideration
in particular photoelectric absorption and Compton scattering
117
Consider the Figure 421 Detector
5
4
Origin of Incoming γ-rayphoton
2
1 3
Electron
Figure 421 A filled Marinelli beaker showing an incoming photon and possibilities of interaction in both the beaker and detector The incoming γ-ray of a particular energy interacts with an electron of the sample in the
Marinelli beaker and there are several possibilities by which it can interact These are
photoelectric absorption (1) and Compton scattering (2) and another Compton scattering (3)
The scattering photon could get deviated again leading to photoelectric absorption in the
detector as in (4) Process (3) is more dominant when the sample is denser while process (2) is
most likely when the sample is less dense The incoming γ-ray photon in (4) will lose some
energy proportional to the density of the sample and thus contributing more to low energy
photons at the Compton continuum This explains the reason why the fit at the region from
50 keV to 300 keV is underestimated at the continuum for sample of higher density such as in
Figure 320 (a) Process (4) explains the overestimation of the lower density sample in Figure
320 (b) Another process is direct photoelectric absorption (5) on a crystal
118
CHAPTER 5
Conclusion This chapter reviews the results of this study and future work on the study is suggested
51 Outlook
This study introduced a novel technique that is the FSA method of analysis to the analysis of
soil spectra using HPGe detector in the ERL at iThemba LABS to complement the
conventional Window Analysis method
Although a correlation was obtained between these two methods of analysis the interpretation
of the results was found to be difficult since the volume of reference 238U and 232Th material
used to obtain standard spectra were only 400 ml whereas the samples to be measured were all
close to the full capacity of Marinelli beakers (volume ~ 900 ml) This resulted in the different
self-absorption effects during the measurement of standard spectra than during the
measurement of sample spectra
Furthermore the difference in volume also leads to the different efficiencies caused by the
change in the geometry For a full Marinelli beaker the average distance from the sample to the
detector is larger than in the 400 ml case In order to avoid the systematic error associated with
this volume effect it is recommended new standard spectra be obtained by measuring
Marinelli beakers filled with reference 238U and 232Th material to obtain constant geometry In
order to accomplish this additional reference material will have to be purchased After these
new standard spectra are obtained the only significant remaining effects that need to be
considered is that associated with density and the effective charge of the material
The model of Sima et al [Sim92] used to calculate the self-absorptions in Marinelli beakers
was found to under-predict the volume effect observed in this work There is therefore scope
to improve this model An alternative to this is to use a Monte Carlo simulation approach
119
The FSA method has shown some difficulty in determining activity concentrations of low
activity samples especially if the background counts are higher than the net counts From
Figures 43 46 49 and 412 in Chapter 4 it is clear that measuring times have to be increased
for WA and FSA short measurements in order to obtain good counting statistics with these
methods The uncertainties increase with a decrease in activity concentrations and this is due
to the contribution of the background counts This can be witnessed in the results of low
activity samples such as the West coast sand sample in Figure 49 where uncertainties
presented for both methods are high
In the FSA method of analysis all the information is included in the spectrum in contrast to
the choosing of regions of interest of prominent peaks in the WA Ignoring the Compton
continuum in the WA simply implies throwing away some counts that can be used for data
analysis
In order to minimise the inevitable self-absorption effects in an attempt to obtain optimal
results in this study the cut-off energy of 300 and 400 keV which gave best least square fit
were therefore chosen
In future if more focus could be put on the absorption effects at low energies for the FSA
method then the accuracy and precision of this technique could improve even further
Perhaps with the help of simulations from software programs such as the Monte Carlo N
Particle extension transport code (MCNPX) which the ERL has at the moment introduced
an effort could be made in the future to understand these effects even better
120
Appendix A
A-1 Some Formulae for combining uncertainties
Type of equation from which
Result R is to be calculated
Formula for calculating the
uncertainty ∆R
R = A plusmn B
∆R = 22 )()( BA ∆+∆
R = a A plusmn b B
Where a and b are constants ∆R= 22 )()( BbAa ∆+∆
R = Ap
Where p is a constant
R = a AP
Where a and p are constants AAP
RR ∆
=∆
R = AP Bq
Where p and q are constants
22
∆
+
∆
=∆
BBq
AAp
RR
AAP
RR ∆
=∆
Table A1 the table shows some of the equations used in the calculation of the uncertainty ∆R derivation from [Kno79]
121
A-2 Means and Standard deviation
The mathematical operation commonly applied to a set of measured or deduced values
( i of a quantity X is to derive an average or mean value ix )21 n= x and an estimate of the
uncertainty of x s( x ) If the values to be averaged have no associated uncertainties the
arithmetic mean is simply given as
sum=
=n
iix
nx
1
1 (A1)
or otherwise the average is calculated as the weighted mean
(A2) sum sum= =
=n
i
n
iiii wxwx
1 1)()(
were is a weighting factor defined as the inverse of the squared standard deviation iw
If the parent distribution is a Gaussian or Poisson distribution and if the sample of the n values
has been obtained from repeated measurements under identical measuring
conditions the arithmetic mean of equation (A1) is the best estimate of the expectation value
m of the parent distribution With increasing sample size the arithmetic mean
ni xxx 2
n x approaches
m[Deb88]
The variance σ2 of the parent ditribution is estimated by the experimental or sample varience
2
1
2 )(1
1)( xxn
xsn
ii minus
minus= sum
=
(A3)
The relative standard deviation s(x) x is also called the variation coefficient υ
122
The experimental variance of the mean s2( x ) denoted by var( x ) is smaller than s2(x) by a
factor 1n ie
)1(
)()()( 1
22
2
minus
minus==
sum=
nn
xx
nxsxs
n
ii
(A4)
Many counting experiments produce only a single result say N counts In these cases we take
N as the estimate of m since m = σ2 for a Poisson distribution N is taken as experimental
standard deviation s(N)
If we have a set of values each of which has an associated variance and if these
variances are not equal the weighted mean defined in (A1) is a better estimate for m than the
arithmetic mean For the weighting factors the reciprocals of the variances are taken ie
ix is 2
ii sw 21=
The variance of x may be derived in two ways The external variance is obtained in analogy to
equation (A3) with weighting factors included ie
sum
sum
=
=
minus
minus= n
ii
n
iii
wn
xxwxs
1
1
2
2
)1(
)()1( (A5)
The internal variance is
sum=
= n
iiw
xs
1
2 1)2( (A6)
123
The magnitude of χ2R the reduced chi-square which is the measure of the goodness of fit to
the data is given by
2
1
2 )(1
1 xxwn i
n
iiR minus
minus= sum
=
χ (A7)
where χR2 is expected to be of the order of unity for a perfect fit to the data [Deb88]
124
Appendix B FORTRAN program
(program used for converting channel numbers to equivalent energy in keV)
c Note that E = a + b Ch or Ch = Eb - ab c c therefore the channel that corresponds to energy i keV c c is given by Ci=ib - ab c and C(i+1) = (i+1)b - ab c Here Ci and Ci+1 are floating point numbers c When we transform to the integer value one will still c get that Ci and Ci+1 may cover two or three channels c c change energy scale c note that n KeV is in ch n+1 dimension eold(5000)ich(5000)countn(5000)cntnew(5000) open(unit=5file=inputfiletxt) open(unit=7file=outputfiledat) do 50 i=17 c read(5) c 50 continue c read livetime read(545)time
45 format (26xle167) read(5)
read(555) a read(555)b read(555)c 55 format (55xle167) c imax=4100 write() imaxabc do 70 k=14 read(5) 70 continue do 200 i=1imax read(5)ichcountn(i) 200 continue write()(iCH(i)eold(i)countn(i)) 300 continue
c now change the energy scale for b=06471 newmax=imaxb+10 do 1000 i=0newmax
125
Epold1=ib - ab Epold2=(i+1)b - ab Else Epold1=(-b+sqrt(bb-4c(a-i)))(20c) Epold2= (-b+sqrt(bb-4c(a-i-1)))20c) endif c j=int(Epold1) jp1=j+1 jp2=j+2 jp3=j+3 jprime=int(Epold2) cntnew(i)=(jp1-Epold1)countn(jp1) c if ((jprime-j)eq1) then cntnew(i)=cntnew(i)+(epold2-jp1)countn(jp2) else c if it spans 3 channels cntnew(i)=cntnew(i)+countn(jp2)+(Epold2-jp2)countn(jp3) endif 1000 continue c print do 2000 i=1newmax write(7)icntnew(i) 2000 continue end
126
Appendix C Background Spectrum
0
0002
0004
0006
0008
001
0012
0014
0016
0018
002
50 550 1050 1550 2050 2550
Energy ( KeV)
Cou
nts
seco
nd
Figure C-1 The background spectrum used in this study It was obtained by measuring a Marinelli beaker filled with tap water
127
Appendix D Self-absorptions effects (by Modelling of the γ-ray transmission through a Marinelli beaker)
According to the model developed by Sima [Sim92] and used here the γ-ray transmission
through a sample-filled Marinelli beaker can be calculated using the expression
teF
t
a micromicro
microminusminus=
1)( (D1)
where F is the transmission factor which is a function of the γ-ray linear attenuation
coefficient and the γ-ray energy t is the effective thickness (of the sample being measured) as
viewed from a small spherical detector This detector is placed in relation to the Marinelli
beaker as shown in Figure D1 The model assumes the detector to be a spherical point
raxis
130 mmD
re ri
0
hi
he
h1
85 mm
θ
H
h2
haxis
73 mm
XS
h0
ri re R
128
r axis132 mm
Figure D1 The Marinelli Beaker with a spherical detector model at the center
The filling of the re-entrant hole he-hi approximately equals the thickness re-ri This thickness
was determined according to the model developed by Sima [Sim92] and the value of this
thickness was recorded for different filling heights These dimensions are tabulated in Table
D1
The effective thickness t can then in principle be determined as follows
int
intΩ
Ω=
d
dl )(θt (D2)
where l(θ) is the thickness of the sample corresponding to the angle θ (see Figure D1) and
denotes integration over the solid angle subtended by the sample
int
Sima showed that t can then be calculated from the expression
t (D3) )]()()()([10201 hrfhrfhrfhrf iiee minusminus+
Ρ=
where (D4)
220
01
2
irh
h
++=
Π∆Ω
=Ρ
(D4)
with
1)ln[(2
)(tan)( 21 ++= minus hrhrhgrhrf ] (D5)
129
with r and h being the dimensions of the model Marinelli beaker shown in Figure D1 The
values of r and h for the Marinelli beaker used here are given in Table D3 For a fully filled
Marinelli beaker the effective thickness t of the sample was found to be 26 cm The effective
thickness for the beaker filled to 500 ml was also calculated (see Table D1)
Volume of sand in Marinelli
beaker (ml)
he= height of sand
Marinelli beaker
dimension (mm)
Effective thickness t (mm)
1000
111
259
500
73
159
Table D1 Results of calculations of the effective thickness t for two Marinelli volumes
γ-ray mass attenuation (microρ )coefficients in various materials can be found in compilation of
Huumlbbel [Hub82] In the use of a generic compound XY where X and Y are two different
elements one can calculate the attenuation coefficient for the sample using the expression
))()(
)())()(
)()YX
Y
YX
XXY ρmicroρmicro
ρmicroρmicroρmicro
ρmicroρmicro
++
+=( (D5)
An example in this case is silicon oxide (SiO2) which is a major constituent of soil We used
the Sima formula to calculate the transmission through a Marinelli beaker filled with water
KCl generic sand 232Th and 238U sources Calculations were made from 50 keV to 2 MeV in
steps of 50 keV The assumed elemental composition of the 238U and 232Th are given in the
130
IAEA manual [RL148] however we assumed the SiO2 to be the major constituent for these
two elements
The calculated transmission factors are shown in Table D2
Water KCl
Density 13 gcm3
Sand
Density 16 gcm3
U (Source)
Density 12 gcm3
Th (source)
Density 13 gcm3
Energy
(keV) Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fwater
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FKCl
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
Fsand
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FU(source)
Mass attenuation
(microρ) in
(cm2g)
Transmission
Factor
FTh(source)
50 02262 0740 07568 0345 03165 0536 03165 0611 03165 0595
100 01707 0794 02196 0693 01682 0704 01682 0760 01682 0749
200 0137 0830 01292 0801 01255 0766 01255 0813 01255 0804
300 01187 0850 01066 0832 01076 0795 01076 0837 01076 0828
500 00969 0875 00852 0862 00874 0828 00874 0864 00874 0857
800 00787 0897 00688 0887 00709 0858 00709 0888 00709 0882
1500 00576 0923 00503 0915 00519 0893 00519 0916 00519 0912
2000 00494 0934 00436 0926 00447 0907 00447 0927 00477 0923
Table D2 A table of the mass attenuation coefficients and the transmission factors
131
Beaker Dimensions
re ri r hi h2 he h1 h0 66 mm
443 mm
48 mm 75 mm 375 mm 111 mm
735 mm 375 mm
Table D3 The dimensions of the full Marinelli Beaker used in the iThemba ERL (for a half full Marinelli (500ml) he = 73 mm)
132
Appendix E Ratios of activity concentrations
Sample type Nuclide Long measurement activity concentration
ratios (WAFSA)
Short measurement activity concentration
ratios (WAFSA) 40K 097 plusmn 005 15 plusmn 02
232Th 16 plusmn 03 14 plusmn 02 West Coast sand
238U 125 plusmn 008 74 plusmn 30 40K 101 plusmn 002 086 plusmn 006
232Th 106 plusmn 006 07 plusmn 01 Westonaria sand
238U 113 plusmn 002 107 plusmn 002 40K 101 plusmn 002 102 plusmn 006
232Th 106 plusmn 004 108 plusmn 01 Kloof mine Sand
238U 118 plusmn 002 117 plusmn 01 40K 097 plusmn 003 100 plusmn 01
232Th 101 plusmn 006 093 plusmn 005 Brick clay
238U 104 plusmn 005 104 plusmn 005 Table E1 The ratios of the activity concentrations (WAFSA) and the associated uncertainties for samples used in the study The ratios are shown for both short and long measurements
133
References [Ame00] Amersham QSA catalogue (2000)
[Bei87] Beiser AConcepts of Modern Physics fourth edition McGraw-Hill Book Co
Singapore (1987)
[Chu94] Chuma JL Physica Reference Manual 4004 Wesbrook Mall Vancouver BC
Canada V6T 2A3 (1994)
[Cre87] Creagh DC The Resolution of Discrepancies in Tables of photon Attenuation
Coefficients Radiation Physics proceedings of the International Symposium on
Radiation Physics University di Ferrara Ferrara Italy (1987)
[Deb88] Debertin K Helmer RG Gamma - and X -Ray Spectroscopy with
semiconductor detectors Elsevier science publishers BV Amsterdam (1988)
[Deb89] Debertin K and Ren JMeasurement of Activity of Radioactive Samples in Marinelli
beakers Nucl Instrum Meth 278 541 (1989)
[Dem97] De Meijer RJ Stapel C Jones DG Roberts PD Rozendaal A MacDonald
WG Improved and New Uses of Radioactivity in Mineral Exploration and Processing
Explor Mining Geology 6 105-117 (1997)
[Dem98] De Meijer RJ Heavy Minerals From Edelstein to Einstein J Geochemical
Exploration 62 81 (1998)
[Dry89] Dryak P Kovar P Plchova L Suran J Correction for the marinelli geometry J
Radioanal Nucl Chem letters 135 281 (1989)
134
[EML97] Environmental Measurements Laboratory Dept of Energy (USA) HASL-30028th
edition (1997)
[Eva55] Evans RDThe Atomic Nucleus McGraw-Hill New York (1955)
[Fel92] Felsmann M Denk HJ Efficiency Calibration of Germanium Detectors with Internal
Standards J RadioanalNucl Chem 157 47 (1992)
[Fuumll99] Fuumlloumlp M Portable gamma spectrometers for environmental and industrial applications at the
institute of preventative and clinical medicine in Bratislava Bratislava Slovakia Industrial
and environmental applications of nuclear techniques report of a workshop held in
Vienna 7-11 September 1998 International Atomic Energy Agency (IAEA)
(1999)
[Gar01] Garcia-Talavera M Laedermann JP Decombaz M Daza MJ Quintana B
Coincidence summing corrections for the natural decay series in γ-ray spectrometry Journal of
Radiation and Isotope 54 769 (2001)
[Gil95] Gilmore G Hemingway JD Practical gamma-ray spectrometry John Wiley amp
Sons New York (1995)
[Hew85] Hewitt PG Conceptual Physics Fifth edition City College of San Francisco (1985)
[Hen01] Hendriks PHGM Limburg J de Meijer RJ Full-spectrum analysis of natural
gamma-ray spectra J Environmental Radioactivity 53 365 (2001)
[Hen03] Hendriks P In-depth γ-ray studies Borehole measurements Rijksuniversiteit
Groningen Published PhD Thesis (2003)
135
[IAE89] International Atomic Energy Agency Measurement of radionuclides in food and
environment (Technical report series No295) (1989)
[ISO03] Draft international standard ISODis 18589-1 Measurement of radioactivity in the
environment-soil-part 1 general guide and definition International Organisation for
Standardisation (2003)
[Jos04] Joseph AD iThemba LABS South Africa (Private communication) (2004)
[Kno79] Knoll GF Radiation detection and measurement John Wiley amp Sons New York
(1979)
[Kno00] Knoll GF Radiation detection and measurement third edition John Wiley amp Sons
New York (2000)
[Kra88] Krane KS Introductory Nuclear Physics John Wiley amp Sons Inc New York 10158
(1988)
[Leo87] William R Leo Techniques for nuclear and particle physics experiments Springer-Verlag
Berlin (1987)
[Lil01] Lilley J Nuclear Physics Principles and Applications John Wiley amp Sons New York
(2001)
[Mal01] Maleka PP Calibration of germanium detector for applications of radiometric
methods in South Africa University of the Western Cape RSA Unpublished MSc
thesis (2001)
136
[Mer97] Merril E Thomas G Environmental Radioactivity from natural industrial and military
sources fourth edition Academic press publishers San Diego (1997)
[Mil94] Miller KM Shebell P Klemic GA In situ gamma-ray spectroscopy for the measurement of
uranium in surface soils Health Physics 67 140 (1994)
[Par95] Park TS Jeon WJ Measurement of Radioactive Samples in Marinelli Beakers by Gamma-
ray Spectroscopy Journal of RadAnal and Nucl Chem 193 133 (1995)
[Pre92] Press WH Teukolsky SA Vetterling WT Flannery BP Numerical Recipes in
Fortran second edition (1992)
[Qui72] Quittner P Gamma-ray spectroscopy with particular reference to detector and computer
evaluation techniques Akademia Kiado Budapest (1972)
[RL148] Preparation and Certification of IAEA Gamma Spectrometry Reference Materials RGU-1
RGTh-1 and RGK-1 AQCS Intercomparison Runs Reference materials IAEA
[Ral72] Ralph E Lapp amp Howard L Andrews Nuclear Radiation Physics fourth edition
Prentice- Hall Inc Englewood Cliffs New Jersey (1972)
[Sha72] Shapiro J Radiation Protection A guide for scientists and physicists Harvard
University press Cambridge Massachusetts (1972)
[Sim92] Sima O Photon Attenuation for samples in Marinelli beaker geometry An analytical
computation Health Physics 62 445 (1992)
137
[Sta97] Stapel C Limburg J de Meijer RJ Calibration of BGO scintillation detectors in a bore
hole geometry Nuclear Geophysics Division (NGD) Kernfysinch Versneller Instituut
(KVI) Rijksuniversiteit Gronigen (1997)
[USR98] Germanium detectors Users Manual Canberra industries 800 Research Parkway
Meriden CT 06450 (1998)
[Van02a] Van Wijngaarden LB Venema RJ De Meijer RJ Zwasman JJG Van Os G
Gieske JMJ Radiometric Sand-mud characterization in the Rhine-Meuse estuary part A
fingerprinting Geomorphology 43 87 (2002)
[Van02b] Van Rooyen TJ Lecture notes The science of radiation protection iThemba LABS
unpublished (2002)
[Ven01] Venema LB de Meijer RJ Natural radionuclides as tracers of the dispersal of dredge spoil
dumped at sea J Environmental Radioactivity 55 221 (2001)
[Ver92] Verplancke J Low-level gamma spectroscopy low lower lowest Nuclear Instruments and
Methods in Physics Research 312 174 (1992)
[www01] wwwscescapenet~woodselementspotassiumhtml taken on 17082004
[www02] wwwdohwagovehprpairFact taken on 4092003
138
- Title Page
- Declaration
- Acknowledgements
- Abstract
- Table of Contents
- List of Tables
- List of Figures
- Chapter 1
-
- 1 Introduction
-
- 11 The atomic nuclei and radioactivity an overview
-
- 111 Radioactive Decay
- 112 Decay modes
- 113 Decay rates
-
- 12 Types of environmental radioactivity
- 13 Review of primordial radioactivity
-
- 131 Decay series of natural radionuclides
-
- 14 Literature review natural radionuclide activity concentration in soil
-
- 141 Methods used to determine activity concentrations in soil
- 142 Interaction of gamma rays with matter
-
- 1421 Photoelectric absorption
- 1422 Compton Scattering
- 1423 Pair Production
- 1424 Attenuation Coefficients
-
- 143 Examples of gamma-ray spectrometry systems
-
- 15 The motivation for this study
- 16 The aim and scope of this study
- 17 Thesis outline
-
- Chapter 2
-
- Experimental Methods
-
- 21 The HPGe gamma-ray detection facility
-
- 211 Physical layout
- 212 HPGe technical specifications
- 213 Electronics
- 214 Figure of merit
-
- 22 Sample Selection Preparation and Measurement
- 23 Reference sources
- 24 Data acquisition
-
- Chapter 3
-
- Data Analysis
-
- 31 Window Analysis
-
- 311 Determination of γ-ray detection efficiency
- 312 Treatment of uncertainties in WA
-
- 32 Full Spectrum Analysis
-
- 321 Derived standard spectra
- 322 Chi-square minimization technique
- 323 Treatment of uncertainties for FSA
-
- Chapter 4
-
- Results and Discussion
-
- 41 WA results
- 42 FSA results
- 43 Comparison of WA and FSA results
- 44 Discussion of WA Results
- 45 Discussion of FSA Results
-
- Chapter 5
-
- Conclusion
-
- 51 Outlook
-
- Appendices
- References
-
- 2005-08-15T134503+0000
- Library
- Document is released